WO2018049843A1 - 三维传感器系统及三维数据获取方法 - Google Patents

三维传感器系统及三维数据获取方法 Download PDF

Info

Publication number
WO2018049843A1
WO2018049843A1 PCT/CN2017/086258 CN2017086258W WO2018049843A1 WO 2018049843 A1 WO2018049843 A1 WO 2018049843A1 CN 2017086258 W CN2017086258 W CN 2017086258W WO 2018049843 A1 WO2018049843 A1 WO 2018049843A1
Authority
WO
WIPO (PCT)
Prior art keywords
dimensional
dimensional point
candidate
line
camera
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2017/086258
Other languages
English (en)
French (fr)
Inventor
郑俊
陈尚俭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Scantech Co Ltd
Original Assignee
Hangzhou Scantech Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hangzhou Scantech Co Ltd filed Critical Hangzhou Scantech Co Ltd
Priority to EP21177443.5A priority Critical patent/EP3907702B1/en
Priority to US15/573,487 priority patent/US10309770B2/en
Priority to KR1020197010317A priority patent/KR102096806B1/ko
Priority to EP17850048.4A priority patent/EP3392831B1/en
Publication of WO2018049843A1 publication Critical patent/WO2018049843A1/zh
Anticipated expiration legal-status Critical
Priority to US16/428,007 priority patent/US11060853B2/en
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
    • G01B11/25—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures by projecting a pattern, e.g. one or more lines, moiré fringes on the object
    • G01B11/2518—Projection by scanning of the object
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/50—Depth or shape recovery
    • G06T7/55—Depth or shape recovery from multiple images
    • G06T7/586—Depth or shape recovery from multiple images from multiple light sources, e.g. photometric stereo
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
    • G01B11/25—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures by projecting a pattern, e.g. one or more lines, moiré fringes on the object
    • G01B11/2545—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures by projecting a pattern, e.g. one or more lines, moiré fringes on the object with one projection direction and several detection directions, e.g. stereo
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/80—Geometric correction
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/97—Determining parameters from multiple pictures
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G06T2207/10012—Stereo images
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30244—Camera pose

Definitions

  • the present invention relates to the field of three-dimensional scanning, and in particular to a three-dimensional sensor system and a three-dimensional data acquisition method.
  • Optical three-dimensional scanning technology is gradually being used in various fields of industry, archaeology, medical treatment, teaching, etc., and the triangulation method is widely used because of its wide application, high accuracy and high cost performance.
  • Products using the principle of triangulation include laser ranging sensors, three-dimensional contour sensors, three-coordinate laser probes, hand-held laser three-dimensional scanners, hand-held projection three-dimensional scanners, and fixed projection three-dimensional scanners.
  • Triangulation Projecting a single line pattern with a laser or pattern projector is a common form of triangulation implementation, and its implementation principle is relatively simple.
  • the camera photosensitive element Capturing a projected line pattern on the image, connecting a point on the projected line pattern to the optical center of the camera, and the intersection of the line and the plane projected by the pattern projector is the desired scan Three-dimensional points on the surface of the object.
  • Triangulation has high scanning accuracy, fast scanning speed, uncomplicated hardware and high cost performance, so it is widely used in close-range non-contact scanning. However, scanners or three-dimensional sensors that generally use triangulation only project a line pattern.
  • the image may be mismatched, and effective three-dimensional data cannot be obtained. Therefore, the speed of the conventional scanner or three-dimensional sensor using the triangulation method is greatly limited, and it is not suitable for some occasions where the scanning speed is high.
  • the present invention aims to provide a three-dimensional sensor system that adopts multiple linear pattern projection methods, has higher exit efficiency and faster scanning speed.
  • the present invention adopts the following technical solutions:
  • a three-dimensional sensor system comprising at least one pattern projector, at least two cameras, a two-dimensional image feature extractor, a three-dimensional point cloud generator, and a three-dimensional point cloud checker;
  • the pattern projector is configured to simultaneously project at least two line patterns
  • the at least two cameras are configured to synchronously capture a two-dimensional image of the scanned object
  • the two-dimensional image feature extractor is configured to extract a two-dimensional line set of the at least two line patterns on the surface of the scanned object on the two-dimensional image;
  • the three-dimensional point cloud generator is configured to generate the set of candidate three-dimensional points by using the two-dimensional line set;
  • the three-dimensional point cloud checker is configured to filter, from the candidate three-dimensional point set, a real three-dimensional point set that correctly matches a projection contour of the surface of the object.
  • the two-dimensional image feature extractor is configured to extract a two-dimensional line set of the at least two line patterns on the surface of the scanned object on the two-dimensional image, and specifically includes: the two-dimensional The image feature extractor corrects the two-dimensional image according to the internal reference of the camera corresponding to the two-dimensional image, and extracts a connected region of the line contour in the corrected image according to the pixel gray difference, and then according to the gray in the connected region
  • the center of gravity calculation obtains a two-dimensional line set of sub-pixel-level highlight centers.
  • the three-dimensional point cloud generator is configured to generate the set of candidate three-dimensional points by using the two-dimensional line set, specifically: the three-dimensional point cloud generator from Extracting two-dimensional point data from the two-dimensional line sets of at least two synchronous two-dimensional images, and using the spatial positional relationship between the cameras, calculating the candidate three-dimensional point set according to the trigonometric principle and the polar line constraint principle ;
  • the three-dimensional point cloud checker is configured to filter, from the candidate three-dimensional point set, a set of real three-dimensional points of the projected contour of the surface of the correct matching object, specifically: the three-dimensional point cloud checker according to the Whether the point in the candidate three-dimensional point set is in a certain three-dimensional light surface projected by the pattern projector to determine whether the point belongs to a real three-dimensional point set, and performs screening to obtain a true three-dimensional point set.
  • the three-dimensional point cloud checker determines whether the point belongs to a real three-dimensional image according to whether a point in the candidate three-dimensional point set is in a certain three-dimensional light surface projected by the pattern projector.
  • Point collection, and screening to obtain a true three-dimensional point set specifically including: the candidate three-dimensional point set includes a plurality of subsets, and the three-dimensional point cloud checker uses the distance of the subset to the three-dimensional smooth surface as According to the method, the sub-set with the smallest distance is selected as the real three-dimensional point set.
  • the number of cameras is at least three;
  • the three-dimensional point cloud generator is configured to generate the candidate three-dimensional point set by using the two-dimensional line set, specifically: the three-dimensional point cloud generator separately extracts two two-dimensional line sets of two synchronous two-dimensional images. Dimension point data, using the spatial positional relationship of the two cameras corresponding to the two synchronous two-dimensional images, and calculating the candidate three-dimensional point set according to the trigonometric principle and the polar line constraint principle;
  • the three-dimensional point cloud checker is configured to filter, from the candidate three-dimensional point set, a set of real three-dimensional points of a projection contour that correctly matches an object surface, and specifically includes: the three-dimensional The point cloud checker performs data verification on the candidate three-dimensional point set by using a two-dimensional image taken by a third or more cameras, and performs screening to obtain a true three-dimensional point set.
  • the three-dimensional point cloud checker performs data verification on the candidate three-dimensional point set by using a two-dimensional image captured by a third or more cameras, and performs screening to obtain a true three-dimensional image.
  • the point set includes: the candidate three-dimensional point set includes a plurality of subsets, and the subset and the optical network connection of the third camera and the two-dimensional image captured by the third camera have an intersection set
  • the three-dimensional point cloud checker uses the distance from the intersection point to the two-dimensional line on the two-dimensional image captured by the third camera as a basis, and selects a subset corresponding to the minimum distance to be true. 3D point collection.
  • the three-dimensional point cloud generator is configured to generate the set of candidate three-dimensional points by using the two-dimensional line set, and specifically includes: the image captured by the three-dimensional point cloud generator from any camera Extracting two-dimensional point data from the two-dimensional line set, using the spatial positional relationship between the plurality of spatial light surfaces projected by the pattern projector and the camera, and obtaining the candidate three-dimensional point set according to a trigonometric principle;
  • the three-dimensional point cloud checker is configured to filter, from the candidate three-dimensional point set, a set of real three-dimensional points of the projection contour that correctly matches the surface of the object, specifically: the three-dimensional point cloud checker will
  • the candidate three-dimensional point set is verified with the image of the other at least one camera, and is filtered to obtain a true three-dimensional point set.
  • the three-dimensional point cloud checker verifies the image of the candidate three-dimensional point and the image of the at least one other camera, and performs screening to obtain a true three-dimensional point set, which specifically includes: Selecting a three-dimensional point set includes several sub-sets, There is a set of intersections between the set of sub-sets and the image of the other at least one camera and the image captured by the other at least one camera, and the three-dimensional point cloud checker gathers the other at least one camera with the intersection Based on the distance of the two-dimensional line on the captured image, the sub-set corresponding to the minimum distance is selected as the real three-dimensional point set.
  • the line pattern is projected by one of the pattern projectors, or a plurality of the pattern projectors are simultaneously projected; the line pattern is a straight line or a curved line.
  • the pattern projector includes a line laser and a DOE beam splitting element, and the line laser splits a plurality of laser line segments through the DOE beam splitting element.
  • the pattern projector includes a projector that directly projects the at least two line patterns.
  • the three-dimensional sensor system includes a synchronization trigger for triggering synchronization of the camera and the pattern projector.
  • a three-dimensional data acquisition method includes the following steps:
  • the pattern projector projects at least two line patterns
  • At least two cameras simultaneously capture a two-dimensional image
  • the extracting the two-dimensional line set of the at least two line patterns on the surface of the scanned object on the two-dimensional image comprises: according to the internal reference of the camera corresponding to the two-dimensional image The dimensional image is corrected for distortion, and the connected region of the contour of the line in the corrected image is extracted according to the difference of the gray level of the pixel, and then the two-dimensional line set of the highlight center of the sub-pixel level is obtained according to the center of gravity of the gray in the connected region.
  • the generating the two-dimensional line set to generate the candidate three-dimensional point set comprises: separately extracting the two-dimensional point data from the two-dimensional line set of the at least two synchronous two-dimensional images, by using the camera The spatial positional relationship between the two, according to the principle of trigonometry and the principle of polar line constraints, the candidate set of three-dimensional points is calculated;
  • Selecting, from the candidate set of three-dimensional points, a set of real three-dimensional points of the projected contours of the surface of the correct matching object specifically comprising: according to whether the point in the set of candidate three-dimensional points is projected by the pattern projector
  • a three-dimensional smooth surface is used to determine whether the point belongs to a real three-dimensional point set, and is filtered to obtain a true three-dimensional point set.
  • the set of three-dimensional points includes: the candidate three-dimensional point set includes a plurality of subsets, and the subset of the minimum set of distances is selected as the true three-dimensional point set based on the distance of the subset to the three-dimensional smooth surface.
  • the number of cameras is at least three;
  • Generating the two-dimensional line set to generate an alternate three-dimensional point set specifically comprising: respectively extracting two-dimensional point data from two-dimensional line sets of two synchronous two-dimensional images, using the two Stepping the spatial position relationship of the two cameras corresponding to the two-dimensional image, and calculating the candidate three-dimensional point set according to the trigonometric principle and the polar line constraint principle;
  • Extracting, from the candidate three-dimensional point set, a true three-dimensional point set of the projection contour of the surface of the correct matching object specifically comprising: using the two-dimensional image captured by the third or more cameras to the candidate three-dimensional point
  • the set performs data validation and is filtered to obtain a true set of 3D points.
  • the two-dimensional image captured by the third or more cameras performs data verification on the candidate three-dimensional point set, and performs screening to obtain a true three-dimensional point set, which specifically includes:
  • the candidate three-dimensional point set includes a plurality of sub-sets, and the sub-set and the optical network connection of the third camera and the two-dimensional image captured by the third camera have an intersection set, and the intersection set Based on the distance of the two-dimensional line on the two-dimensional image captured by the third camera, the sub-set corresponding to the minimum distance is selected as a true three-dimensional point set.
  • the generating the two-dimensional line set to generate the candidate three-dimensional point set comprises: extracting two-dimensional point data from the two-dimensional line set of the image captured by any camera, and using the pattern projection The spatial positional relationship between the plurality of spatial light planes projected by the device and the camera, and the candidate three-dimensional point set is obtained according to the trigonometric principle;
  • Selecting, from the candidate three-dimensional point set, the true three-dimensional point set of the projection contour that correctly matches the surface of the object specifically: performing verification and screening the image of the candidate three-dimensional point and the image of the at least one other camera Get a true collection of 3D points.
  • the candidate three-dimensional point set and the image of the at least one other camera are verified, and the real three-dimensional point set is obtained by filtering.
  • the method includes: the candidate three-dimensional point set includes a plurality of subsets, and the subset and the image of the other at least one camera are connected to the image captured by the at least one camera, and the intersection set is To the distance of the two-dimensional line on the image captured by the at least one camera, the sub-set corresponding to the minimum distance is selected as a true three-dimensional point set.
  • the three-dimensional sensor system adopts a plurality of linear pattern projection modes, the three-dimensional sensor system can recognize a plurality of linear patterns projected at the same time, and calculate three-dimensional point cloud data of the surface of the object, and the efficiency of the out point is Several times the traditional single-line scanning, the scanning speed is significantly improved.
  • FIG. 1 is a schematic diagram of a three-dimensional sensor system according to an embodiment of the present invention.
  • FIG. 2 is a schematic diagram of pattern projection and image capture of a three-dimensional sensor system according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of a method for acquiring a real three-dimensional point set according to Embodiment 1 of the present invention
  • FIG. 4 is a schematic diagram of a method for acquiring a real three-dimensional point set in Embodiment 2 of the present invention.
  • FIG. 5 is a schematic diagram of a method for acquiring a real three-dimensional point set according to Embodiment 3 of the present invention.
  • Marking description 100, object; 210, pattern projector; 220, first camera; 221, first image; 230, second camera; 231, second image; 240, third camera; 241, third image; , pattern projector; 320, first camera; 321, first image; 330, second camera; 331, second image; 410, first camera; 411, first image; 420, second camera; Two images; 430, a third camera; 431, a third image; 510, a pattern projector; 520, a first camera; 521, a first image; 530, a second camera; 531, a second image.
  • FIG. 1 is a schematic diagram of a three-dimensional sensor system according to an embodiment of the present invention.
  • the three-dimensional sensor includes a first camera, a second camera, a third camera, a synchronization trigger, a pattern projector, a two-dimensional image extractor, and a three-dimensional point cloud generator. It should be noted that the number of cameras is at least two, and the pattern The number of projectors is at least one, which is not limited herein.
  • the main workflow of the 3D sensor system is as follows:
  • Step 1 The pattern projector projects at least two line patterns.
  • the line pattern may be projected by one of the pattern projectors, or a plurality of the pattern projectors may be simultaneously projected; the line pattern is a straight line or a curved line.
  • the pattern projector includes a line laser and a DOE beam splitting element, the line laser splitting a plurality of laser line segments through the DOE beam splitting element.
  • the pattern projector may also be a projector, and the projector directly projects the at least two line patterns.
  • Step 2 At least two cameras synchronously capture a two-dimensional image of the scanned object.
  • the steps 1 and 2 can be performed simultaneously.
  • the first trigger and the second camera can be triggered by the synchronization trigger while the pattern projector is triggered, and the two frames captured by the two cameras are captured.
  • the feature extraction is performed separately to the two-dimensional image extractor.
  • Step 3 The two-dimensional image feature extractor extracts a two-dimensional line set of the at least two line patterns on the surface of the scanned object on the two-dimensional image.
  • the two-dimensional image extractor performs distortion correction on the image according to the internal reference of the camera corresponding to the two-dimensional image, and extracts a connected region of the high-value gray line contour in the corrected image according to the pixel gray difference, and then according to the region
  • the gray center of gravity calculation obtains a sub-pixel-level highlight center two-dimensional line set, and outputs the obtained two-bit line set to the three-dimensional point cloud generator;
  • Step 4 The three-dimensional point cloud generator generates the candidate three-dimensional point set by using the two-dimensional line set;
  • Step 5 The three-dimensional point cloud checker selects a set of real three-dimensional points that correctly match the surface projection contour of the object from the candidate three-dimensional point set.
  • step 4 and step 5 can be implemented by the following three methods:
  • the first method is: the three-dimensional point cloud generator separately extracts two-dimensional point data from two-dimensional line sets of at least two synchronous two-dimensional images, and uses the spatial positional relationship between the cameras according to the principle of trigonometry and The polar line constraint principle calculates the candidate three-dimensional point set; the three-dimensional point cloud checker determines whether the point in the candidate three-dimensional point set is in a certain three-dimensional light surface projected by the pattern projector Determine whether the point belongs to a real three-dimensional point set, and filter to obtain a true three-dimensional point set.
  • the candidate three-dimensional point set includes a plurality of subsets, and the three-dimensional point cloud checker uses the distance of the subset to the three-dimensional smooth surface as a basis, and the subset with the smallest distance is selected to be true. 3D point collection.
  • the second method is: the three-dimensional point cloud generator separately extracts two-dimensional point data from two sets of two-dimensional lines of the synchronous two-dimensional image, and uses the spatial positions of the two cameras corresponding to the two synchronous two-dimensional images. Relationship, and calculating the candidate three-dimensional point set according to the trigonometric principle and the polar line constraint principle; the three-dimensional point cloud checker uses the two-dimensional image taken by the third or more cameras to prepare the preparation A three-dimensional point set is selected for data verification, and a real three-dimensional point set is obtained by screening.
  • the candidate three-dimensional point set includes a plurality of subsets, and the subset and the optical network connection of the third camera and the two-dimensional image captured by the third camera have an intersection set,
  • the three-dimensional point cloud checker uses the distance from the intersection point to the two-dimensional line on the two-dimensional image captured by the third camera as a basis, and selects a subset corresponding to the minimum distance to be a true three-dimensional point set. .
  • a third method is: the three-dimensional point cloud generator extracts two-dimensional point data from the two-dimensional line set of images captured by any camera, and uses the plurality of spatial light surfaces projected by the pattern projector and the a spatial positional relationship of the camera, the candidate three-dimensional point set is obtained according to a trigonometric principle; the three-dimensional point cloud checker verifies and selects the image of the candidate three-dimensional point and the image of the at least one other camera Get a true collection of 3D points.
  • the candidate three-dimensional point set includes a plurality of subsets, and the subset of the optical center of the subset and the at least one camera has an intersection set with the image captured by the at least one camera, the three-dimensional point a cloud checker gathers the said points at said intersection
  • the distance of the two-dimensional line on the image captured by at least one camera is used as a basis, and the subset corresponding to the minimum distance is selected as a true three-dimensional point set.
  • FIG. 2 is a schematic diagram of pattern projection and image capture of a three-dimensional sensor system according to an embodiment of the present invention.
  • a camera with a known mutual spatial positional relationship and a pattern projector capable of simultaneously emitting three linear patterns are taken as an example.
  • the three-dimensional sensor system includes a pattern projector 210, a first camera 220, and a first The image 221, the second camera 230, the second image 231, the third camera 240, and the third image 241.
  • the steps to generate a two-dimensional line set are as follows:
  • the pattern projector 210 projects three smooth surfaces PL 1 , PL 2 and PL 3 , and the smooth surface forms three three-dimensional spatial lines SR 1 , SR 2 and SR 3 on the surface of the object 100, the first camera 220 and The second camera 230 synchronously captures the two-dimensional image, and the first camera 220 and the second camera 230 respectively capture the first image 221 and the second image 231, and the first image 221 and the second image 231 include a part of the surface of the object 100.
  • a line pattern, the line pattern being presented in the form of a two-dimensional line on the two-dimensional figure, SA 1 , SA 2 , SA 3 and SB 1 , SB 2 , SB 3 , respectively .
  • three methods for generating an alternative three-dimensional point set and extracting a real three-dimensional point set from the candidate three-dimensional point set are respectively described in the following three specific embodiments.
  • the following three implementations are respectively described.
  • the examples are all based on the two-dimensional line set obtained by the three-dimensional sensor system shown in FIG. 2, and respectively correspond to the foregoing three methods for acquiring a real three-dimensional point set.
  • FIG. 3 is a schematic diagram of a method for acquiring a real three-dimensional point set according to Embodiment 1 of the present invention.
  • the schematic diagram includes a pattern projector 310, a first camera 320, a first image 321, a second camera 330, and a second image 331.
  • the pattern projector 310 projects three smooth surfaces PL 1 , PL 2 and PL 3 , the first camera 320 and the second camera 330 synchronously capture a two-dimensional image, and the first camera 320 and the second camera 330 respectively capture the first image.
  • 321 and the second image 331, O 1 , O 2 are the optical centers of the first camera 320 and the second camera 330, respectively, and the internal and external parameters of the camera are known.
  • the method includes the following steps:
  • the first image 321 and the second image 331 include a plurality of linear patterns on a part of the surface of the object 100, and the line patterns are presented in the form of two-dimensional lines on the two-dimensional graphics, such as the first image.
  • 321 two-dimensional line SA 1, Pa i is a two-dimensional two-dimensional point on the line SA 1; a first camera 320 of the second internal reference M a and M B the known internal reference camera 330, a first camera 320 relative to the second camera
  • the method for calculating the three-dimensional point coordinates is as follows: the first image 321 captured by the first camera 320 is imaged on the photosensitive element plane PF 1 , and the second image 331 captured by the second camera 330 is imaged on the photosensitive element plane PF 2 ;
  • the point Pa i on the SA 1 is spatially connected to the optical center O 1 of the first camera 320 as L 1 , and the point Pb 1i on the two-dimensional line SB 1 on the second image 331 and the optical center O 2 of the second camera 330
  • the spatial connection is L 2
  • the intersection of L 1 and L 2 at the spatial point P 1i is one of the candidate three-dimensional points corresponding to the Pa i point; if the spatial lines do not intersect, the two vertical lines are used.
  • 1 ⁇ i ⁇ n ⁇ on the two-dimensional line SA 1 on the first image 321 are ⁇ P 1i
  • the minimum W m min (W k
  • k 1,2,3) obtained by the screening, that is, the determination P ji ⁇ PL m , that is, the set of three-dimensional points ⁇ Pm i
  • the real three-dimensional point set corresponding to the two-dimensional line SA 1 on the first image 321 that is, the true three-dimensional contour line projected by the smooth surface PL m onto the surface of the object 100 is imaged on the first camera 320 as a two-dimensional line SA 1 , in the second
  • the image on camera 330 is a two-dimensional line SB j .
  • FIG. 4 is a schematic diagram of a method for obtaining a real three-dimensional point set in Embodiment 2 of the present invention.
  • the schematic diagram includes a first camera 410, a first image 411, a second camera 420, a second image 421, a third camera 430, a third image 431, and a pattern projector (not shown).
  • the two-dimensional image captured by the first camera 410 is the first image 411
  • the two-dimensional image captured by the second camera 420 is the second image 421
  • the two-dimensional image captured by the third camera 430 is the third image 431.
  • a first captured image 411 imaged on the photosensitive element plane PF 1 a second camera 420 of the second captured image 421 imaged on the photosensitive member plane PF 2
  • the third camera 430 captures a third image 431 imaged in a photosensitive element
  • O 1 , O 2 , and O 3 are the optical centers of the first camera 410, the second camera 420, and the third camera 430, respectively, and the internal and external parameters of the camera are known. .
  • the third or more cameras can be used for data verification.
  • the system can verify the spatial positional relationship between the plurality of smooth surfaces projected by the camera and the pattern projector in advance, and use the third camera to verify the three-dimensional points in the obtained candidate three-dimensional point set, and output the true three-dimensional image. Point collection.
  • the intersection of the optical center O 3 of the camera 430 and the third image 431 on the photosensitive element PF 3 is ⁇ Pc 1i
  • the positional relationship between each of the subsets of the intersection points and the three two-dimensional lines ⁇ SC 1 , SC 2 , SC 3 ⁇ on the third image 431 is separately counted.
  • the sum of the distances of each point in the statistical sub-set to a two-dimensional line SC k is used as a criterion for screening:
  • 1 ⁇ i ⁇ n ⁇ on the two-dimensional line SA 1 on the first image 411 is calculated after the polar line (N 3 is the pole) according to the principle of the polar line constraint, the polar line and the two-dimensional line
  • the point set of the candidate three-dimensional point corresponding to the intersection set of SB j is P ji and the optical center O 3 of the third camera 430 intersects with the third image 431 is ⁇ Pc ji
  • 1 ⁇ i ⁇ n ⁇ to the two-dimensional line SC k is D(SC k , Pc ji ).
  • the minimum W m min(W k
  • k 1,2,3) is obtained by screening, that is, the two-dimensional line SC m is a set of real three-dimensional points corresponding to the two-dimensional line SA 1 ⁇ P mi
  • FIG. 5 is a schematic diagram of a method for obtaining a real three-dimensional point set according to Embodiment 3 of the present invention.
  • the schematic diagram includes a pattern projector 510, a first camera 520, a first image 521, a second camera 530, and a second image 531.
  • the pattern projector 510 projects three smooth surfaces PL 1 , PL 2 , and PL 3 .
  • the two-dimensional image captured by the first camera 520 is the first image 521
  • the two-dimensional image captured by the second camera 530 is the second image 531 .
  • the first image 521 captured by the first camera 520 is formed on the photosensitive element plane PF 1
  • the second image 531 captured by the second camera 530 is formed on the photosensitive element plane PF 2
  • O 1 and O 2 are respectively the first
  • the optical centers of the camera 520 and the second camera 530 are known, and the internal and external parameters of the camera are known.
  • step S2 Obtain a two-dimensional point set ⁇ Pa i
  • the extension of the line connecting the heart O 1 intersects the three smooth faces PL 1 , PL 2 and PL 3 at ⁇ P 1i
  • the sum of the distances of each point in the subset to a two-dimensional line SB k is used as a criterion for screening:
  • the three-dimensional surface is associated with the point j PL is set to ⁇ P ji
  • 1 ⁇ i ⁇ n ⁇ associated with the second image 531 to the two-dimensional line SB k is D (SB k , Pb ji ).
  • the minimum W m min (W k
  • k 1, 2,3) obtained by the screening, that is, the SB m is determined as the imaging line of the smooth surface PL j on the second camera 530, that is, the three-dimensional point on the smooth surface PL j
  • 1 ⁇ i ⁇ n ⁇ is a set of real three-dimensional points corresponding to the two-dimensional line SA 1 of the first image 521, that is, the real three-dimensional point set of the smooth surface PLj projected onto the surface of the object 100 is on the first camera 520.
  • the image is a two-dimensional line SA 1
  • the efficiency of the out-point is several times that of the conventional single-line scanning, which significantly improves the scanning speed.
  • the embodiment of the invention further provides a three-dimensional data acquisition method, the method comprising the following steps:
  • Step 101 The pattern projector projects at least two line patterns
  • the pattern projector includes a line laser and a DOE beam splitting element, the line laser splitting a plurality of laser line segments through the DOE beam splitting element.
  • the pattern projector may also be a projector, and the projector directly projects the at least two line patterns.
  • Step 102 At least two cameras synchronously capture a two-dimensional image
  • Step 103 Extract a two-dimensional line set linearly projected on a surface of the scanned object on the two-dimensional image
  • the extracting the two-dimensional line set linearly projected on the surface of the scanned object on the two-dimensional image comprises: correcting the two-dimensional image according to an internal parameter of the camera corresponding to the two-dimensional image, and The connected region of the line contour in the corrected image is extracted according to the pixel grayscale difference, and then the sub-pixel-level highlight center two-dimensional line set is obtained according to the gray center of gravity in the connected region.
  • Step 104 Generate the two-dimensional line set to generate an alternate three-dimensional point set.
  • Step 105 Select a real three-dimensional point set that correctly matches the surface projection contour of the object from the candidate three-dimensional point set.
  • step 104 and step 105 can be implemented by the following three methods:
  • the first method is: extracting two-dimensional point data from two-dimensional line sets of at least two synchronous two-dimensional images, and calculating the spatial positional relationship between the cameras according to the principle of trigonometry and the principle of polar line constraints.
  • the candidate three-dimensional point set ; according to whether the point in the candidate three-dimensional point set is in a certain three-dimensional light surface projected by the pattern projector Determine whether the point belongs to a real three-dimensional point set, and filter to obtain a true three-dimensional point set.
  • the candidate three-dimensional point set includes a plurality of sub-sets, and the sub-set with the smallest distance is selected as a real three-dimensional point set based on the distance of the sub-set to the three-dimensional smooth surface.
  • the second method is: extracting two-dimensional point data from two sets of two-dimensional lines of synchronous two-dimensional images, and utilizing the spatial positional relationship of the two cameras corresponding to the two synchronous two-dimensional images, and according to the principle of trigonometry Calculating the candidate three-dimensional point set by using the principle of the polar line constraint; performing data verification on the candidate three-dimensional point set by using the two-dimensional image captured by the third or more cameras, and performing screening to obtain a true 3D point collection.
  • the candidate three-dimensional point set includes a plurality of subsets, and the subset is connected to the optical network of the third camera and the two-dimensional image captured by the third camera has a set of intersections.
  • the intersection point is set to the distance of the two-dimensional line on the two-dimensional image captured by the third camera as a basis, and the subset corresponding to the minimum distance is selected as the real three-dimensional point set.
  • the third method is: extracting two-dimensional point data from the two-dimensional line set of images captured by any camera, and using the spatial positional relationship between the plurality of spatial light surfaces projected by the pattern projector and the camera, according to The trigonometric principle obtains the candidate three-dimensional point set; the candidate three-dimensional point set is verified with the image of the at least one other camera, and is filtered to obtain a true three-dimensional point set.
  • the candidate three-dimensional point set includes a plurality of subsets, and the subset is connected to the optical center of the other at least one camera and the other at least one camera
  • the captured image has a set of intersection points, and the sub-set corresponding to the minimum distance is selected as a true three-dimensional point set based on the distance of the intersection point to the two-dimensional line on the image captured by the other at least one camera.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Length Measuring Devices By Optical Means (AREA)
  • Geometry (AREA)

Abstract

本发明涉及三维扫描领域，提供了一种三维传感器系统及三维数据获取方法，该三维传感器系统包括：至少一图案投影器、至少两摄像头、一二维图像特征提取器、一三维点云生成器、一三维点云校验器；所述图案投影器，用于同时投射出至少两条线状图案；所述至少两摄像头，用于同步捕捉被扫描物体的二维图像；所述二维图像特征提取器，用于提取所述二维图像上被扫描物体表面所述至少两条线状图案的二维线条集合；所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合；所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，本发明的三维传感器扫描速度更快，出点效率更高。

Description

三维传感器系统及三维数据获取方法 技术领域
本发明涉及三维扫描领域，特别涉及一种三维传感器系统及三维数据获取方法。
背景技术
光学三维扫描技术目前正逐步被用在各种工业、考古、医疗、教学等领域，而其中的三角测量法由于适用面广、准确度高、性价比高而被广泛使用。利用三角测量法原理的产品有激光测距传感器、三维轮廓传感器、三坐标仪激光测头、手持激光三维扫描仪、手持投影式三维扫描仪、固定投影式三维扫描仪等。
用激光或者图案投影器投射出单条线状图案是一种常见的三角测量法实现形式，其实现原理相对简单，在图案投影器发射的光面与摄像头的位置已知的前提下，摄像头感光元件捕捉到图像上的投影线状图案，将所述投影线状图案上的点与摄像头的光心连线，所述连线与图案投影器所投射的光面的交点即为所求的被扫描物体表面三维点。三角测量法扫描精度较高、扫描速度快、硬件不复杂、性价比较高，因而被广泛应用于近距离非接触式扫描的场合。但一般使用三角测量法的扫描仪或三维传感器只投射一条线状图案，如果同时投射出多条线状图案会导致图像的误匹配，从而无法得到有效的三维数据。因而传统的采用三角测量法的扫描仪或三维传感器的出点速度会受到很大的限制，无法适应一些对扫描速度要求较高的场合。
发明内容
针对现有技术的不足，本发明旨在提供一种采用多条线状图案投射方式，出点效率更高，扫描速度更快的三维传感器系统。
为实现上述目的，本发明采用如下技术方案：
一种三维传感器系统，包括至少一图案投影器、至少两摄像头、一二维图像特征提取器、一三维点云生成器、一三维点云校验器；
所述图案投影器，用于同时投射出至少两条线状图案；
所述至少两摄像头，用于同步捕捉被扫描物体的二维图像；
所述二维图像特征提取器，用于提取所述二维图像上被扫描物体表面所述至少两条线状图案的二维线条集合；
所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合；
所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合。
作为本发明的进一步改进，所述二维图像特征提取器，用于提取所述二维图像上被扫描物体表面所述至少两条线状图案的二维线条集合，具体包括：所述二维图像特征提取器根据所述二维图像所对应摄像头的内参对所述二维图像进行畸变矫正，并根据像素灰度差异提取矫正图像中线条轮廓的连通区域，再根据所述连通区域内的灰度重心计算获得亚像素级的高光中心二维线条集合。
作为本发明的进一步改进，所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合，具体包括：所述三维点云生成器从 至少两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述摄像头之间的空间位置关系，根据三角法原理和极线约束原理计算得出所述备选三维点集合；
所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：所述三维点云校验器根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合。
作为本发明的进一步改进，所述三维点云校验器根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所述三维点云校验器以所述子集合到所述三维光面的距离作为依据，筛选出距离最小的子集合即为真实的三维点集合。
作为本发明的进一步改进，所述摄像头数量为至少三个；
所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合，具体包括：所述三维点云生成器从两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述两幅同步二维图像对应的两个摄像头的空间位置关系，并根据三角法原理和极线约束原理计算得出所述备选三维点集合；
所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：所述三维 点云校验器利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合。
作为本发明的进一步改进，所述三维点云校验器利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所述子集合与所述第三个摄像头的光心连线与所述第三个摄像头所拍摄的二维图像存在交点集合，所述三维点云校验器以所述交点集合到所述第三个摄像头所拍摄的二维图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
作为本发明的进一步改进，所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合，具体包括：所述三维点云生成器从任一摄像头所捕捉图像的所述二维线条集合中提取二维点数据，利用所述图案投影器所投射的多个空间光面与该摄像头的空间位置关系，根据三角法原理得出所述备选三维点集合；
所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：所述三维点云校验器将所述备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合。
作为本发明的进一步改进，所述三维点云校验器将所述备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所 述子集合与所述另外至少一个摄像头的光心连线与所述另外至少一个摄像头所拍摄的图像存在交点集合，所述三维点云校验器以所述交点集合到所述另外至少一个摄像头所拍摄的图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
作为本发明的进一步改进，所述线状图案为一个所述图案投影器投射，或多个所述图案投影器同时投射；所述线状图案为直线或曲线线条。
作为本发明的进一步改进，所述图案投影器包括一线状激光器和一DOE分束元件，所述线状激光器通过所述DOE分束元件分出多条激光线段。
作为本发明的进一步改进，所述图案投影器包括一投影仪，所述投影仪直接投出所述至少两条线状图案。
作为本发明的进一步改进，所述三维传感器系统包括一同步触发器，所述同步触发器用于触发所述摄像头和所述图案投影器进行同步工作。
一种三维数据获取方法，包括以下步骤：
图案投影器投射出至少两条线状图案；
至少两摄像头同步捕捉二维图像；
提取所述二维图像上被扫描物体表面至少两条线状图案的二维线条集合；
将所述二维线条集合生成备选三维点集合；
从所述备选三维点集合中筛选出正确匹配物体表面投影轮廓线 的真实三维点集合。
作为本发明的进一步改进，所述提取所述二维图像上被扫描物体表面至少两条线状图案的二维线条集合，具体包括：根据所述二维图像所对应摄像头的内参对所述二维图像进行畸变矫正，并根据像素灰度差异提取矫正图像中线条轮廓的连通区域，再根据所述连通区域内的灰度重心计算获得亚像素级的高光中心二维线条集合。
作为本发明的进一步改进，将所述二维线条集合生成备选三维点集合，具体包括：从至少两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述摄像头之间的空间位置关系，根据三角法原理和极线约束原理计算得出所述备选三维点集合；
从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合。
作为本发明的进一步改进，所述根据备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，以所述子集合到所述三维光面的距离作为依据，筛选出距离最小的子集合即为真实的三维点集合。
作为本发明的进一步改进，所述摄像头数量为至少三；
将所述二维线条集合生成备选三维点集合，具体包括：从两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述两幅同 步二维图像对应的两个摄像头的空间位置关系，并根据三角法原理和极线约束原理计算得出所述备选三维点集合；
从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合。
作为本发明的进一步改进，所述利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所述子集合与所述第三个摄像头的光心连线与所述第三个摄像头所拍摄的二维图像存在交点集合，以所述交点集合到所述第三个摄像头所拍摄的二维图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
作为本发明的进一步改进，将所述二维线条集合生成备选三维点集合，具体包括：从任一摄像头所捕捉图像的所述二维线条集合中提取二维点数据，利用所述图案投影器所投射的多个空间光面与该摄像头的空间位置关系，根据三角法原理得出所述备选三维点集合；
从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：将所述备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合。
作为本发明的进一步改进，所述将备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合，具体 包括：所述备选三维点集合包括若干子集合，所述子集合与所述另外至少一个摄像头的光心连线与所述另外至少一个摄像头所拍摄的图像存在交点集合，以所述交点集合到所述另外至少一个摄像头所拍摄的图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
相比于现有技术，本发明的有益效果在于：
由于所述三维传感器系统采用了多条线状图案投射方式，所述三维传感器系统可以识别出同时投射的多条线状图案，并计算获得物体表面的三维点云数据，其出点的效率是传统单线扫描的数倍，显著的提升了扫描的速度。
上述说明仅是本发明技术方案的概述，为了能够更清楚了解本发明的技术手段，而可依照说明书的内容予以实施，并且为了让本发明的上述和其他目的、特征和优点能够更明显易懂，以下特举较佳实施例，并配合附图，详细说明如下。
附图说明
图1是本发明实施例提供的三维传感器系统示意图；
图2是本发明实施例三维传感器系统图案投影和图像捕捉示意图；
图3是本发明实施例一中真实三维点集合获取方法示意图；
图4是本发明实施例二中真实三维点集合获取方法示意图；
图5是本发明实施例三中真实三维点集合获取方法示意图。
标记说明：100、物体；210、图案投影器；220、第一摄像头；221、第一图像；230、第二摄像头；231、第二图像；240、第三摄像头；241、第三图像；310、图案投影器；320、第一摄像头；321、第一图像；330、第二摄像头；331、第二图像；410、第一摄像头；411、第一图像；420、第二摄像头；421、第二图像；430、第三摄像头；431、第三图像；510、图案投影器；520、第一摄像头；521、第一图像；530、第二摄像头；531、第二图像。
具体实施方式
下面，结合附图以及具体实施方式，对本发明做进一步描述：
图1是本发明实施例提供的三维传感器系统示意图。所述三维传感器包括第一摄像头、第二摄像头、第三摄像头、同步触发器、图案投影器、二维图像提取器、三维点云生成器，需要说明的是摄像头的数量至少为2个，图案投影器的数量至少为1个，在此不做限定。该三维传感器系统的其主要的工作流程如下：
步骤1、所述图案投影器投射出至少两条线状图案。
优选的，所述线状图案可以为一个所述图案投影器投射，或多个所述图案投影器同时投射；所述线状图案为直线或曲线线条。所述图案投影器包括一线状激光器和一DOE分束元件，所述线状激光器通过所述DOE分束元件分出多条激光线段。
优选的，所述图案投影器也可以为一投影仪，所述投影仪直接投出所述至少两条线状图案。
步骤2、至少两摄像头同步捕捉被扫描物体的二维图像。
优选的，步骤1和步骤2可以同时进行，具体的，可以通过所述同步触发器在触发所述图案投影器的同时触发第一摄像头和第二摄像头曝光，两个摄像头所捕捉的两帧图像分别输出到所述二维图像提取器中进行特征提取。
步骤3、所述二维图像特征提取器提取所述二维图像上被扫描物体表面所述至少两条线状图案的二维线条集合。
优选的，所述二维图像提取器根据所述二维图像所对应摄像头的内参对图像进行畸变矫正后，根据像素灰度差异提取矫正图像中高值灰度线条轮廓的连通区域，再根据区域内的灰度重心计算获得亚像素级的高光中心二维线条集合，并将得到的二位线条集合输出至三维点云生成器；
步骤4、所述三维点云生成器将所述二维线条集合生成备选三维点集合；
步骤5、所述三维点云校验器从所述备选三维点集合中筛选出正确匹配物体表面投影轮廓线的真实三维点集合。
具体的，步骤4和步骤5可以通过以下3种方法实现：
第一种方法为：所述三维点云生成器从至少两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述摄像头之间的空间位置关系，根据三角法原理和极线约束原理计算得出所述备选三维点集合；所述三维点云校验器根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合。
优选的，所述备选三维点集合包括若干子集合，所述三维点云校验器以所述子集合到所述三维光面的距离作为依据，筛选出距离最小的子集合即为真实的三维点集合。
第二种方法为：所述三维点云生成器从两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述两幅同步二维图像对应的两个摄像头的空间位置关系，并根据三角法原理和极线约束原理计算得出所述备选三维点集合；所述三维点云校验器利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合。
优选的，所述备选三维点集合包括若干子集合，所述子集合与所述第三个摄像头的光心连线与所述第三个摄像头所拍摄的二维图像存在交点集合，所述三维点云校验器以所述交点集合到所述第三个摄像头所拍摄的二维图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
第三种方法为：所述三维点云生成器从任一摄像头所捕捉图像的所述二维线条集合中提取二维点数据，利用所述图案投影器所投射的多个空间光面与该摄像头的空间位置关系，根据三角法原理得出所述备选三维点集合；所述三维点云校验器将所述备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合。
优选的，所述备选三维点集合包括若干子集合，所述子集合与所述另外至少一个摄像头的光心连线与所述另外至少一个摄像头所拍摄的图像存在交点集合，所述三维点云校验器以所述交点集合到所述 另外至少一个摄像头所拍摄的图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
为了更好的理解上述3种方法，现举例说明。首先参照图2，图2是本发明实施例三维传感器系统图案投影和图像捕捉示意图。图2中以三个相互空间位置关系已知的摄像头和一个可同时发出三条线状图案的图案投影器为例，具体的，该三维传感器系统包括图案投影器210、第一摄像头220、第一图像221、第二摄像头230、第二图像231、第三摄像头240、第三图像241。具体生成二维线条集合的步骤如下：
S1、所述图案投影器210投射出三个光面PL1、PL2和PL3，所述光面在物体100表面形成三条三维空间线条SR1、SR2和SR3，第一摄像头220和第二摄像头230同步捕捉二维图像，第一摄像头220和第二摄像头230分别捕捉第一图像221和第二图像231，第一图像221和第二图像231中包含所述物体100表面一部分的多条线状图案，所述线状图案在所述二维图形上以二维线条的形式呈现，分别为SA1、SA2、SA3和SB1、SB2、SB3。
S2、从第一图像221和第二图像231中分别提取所述图像上所有二维线条SA1、SA2、SA3和SB1、SB2、SB3，所述摄像头的内参已知，先根据对应摄像头内参进行图像畸变矫正，根据对图像进行列遍历统计像素点的灰度的差异，提取图像高值灰度线条轮廓的连通区域，再根据区域内的灰度重心计算获得亚像素级的高光中心二维线条，得到二维线条集合。
下面通过以下3个具体实施例来分别说明前述将所述二维线条集合生成备选三维点集合以及从所述备选三维点集合中筛选出真实三维点集合的3种方法，以下3个实施例都是基于图2所示的三维传感器系统得出的二维线条集合后进行的，且分别对应前述3种获取真实三维点集合的方法。
图3是本发明实施例一中真实三维点集合获取方法示意图。所述示意图包括图案投影器310、第一摄像头320、第一图像321、第二摄像头330、第二图像331。所述图案投影器310投射出三个光面PL1、PL2和PL3，第一摄像头320和第二摄像头330同步捕捉二维图像，第一摄像头320和第二摄像头330分别捕捉第一图像321和第二图像331，O1、O2分别为第一摄像头320和第二摄像头330的光心，所述摄像头的内外参已知。
所述方法包括以下步骤：
3.1、第一图像321和第二图像331中包含所述物体100表面一部分的多条线状图案，所述线状图案在所述二维图形上以二维线条的形式呈现，如第一图像321中二维线条SA1，Pai为二维线条SA1上一个二维点；第一摄像头320的内参MA和第二摄像头330的内参MB已知，第一摄像头320相对第二摄像头330的外参RT已知，根据第一摄像头320相对第二摄像头330的外参RT(其中R为旋转矩阵和T为平移向量)计算出本征矩阵(Essential Matrix)E＝RS，其中
Figure PCTCN2017086258-appb-000001
再利用两个摄像头的内参MA和MB得到基础矩 阵(Fundamental Matrix)F＝(MA -1)TE(MB -1),根据极线约束原理，满足(Xai,Yai,1)TF(x,y,1)＝0，其中(Xai,Yai)为点Pai的位置坐标，(x,y)为第一图像321上极线上点的二维坐标值；得出极线后(N1、N2为极点)，便可以求出其与第二图像331上的所有二维线条的交点的集合{Pb1i,Pb2i,Pb3i}。
3.2、第一图像320上的二维点Pai与交点集合{Pb1i,Pb2i,Pb3i}中的每个点通过三角法原理计算出三维点(鉴于三角法原理是公知常识，这里不再赘述)，这些三维点即为该二维点Pai对应的所有可能的三维点集合，即为备选三维点集合Mi＝{P1i，P2i，P3i}。计算三维点坐标方法如下：第一摄像头320所捕捉的第一图像321成像在感光元件平面PF1上，第二摄像头330所捕捉的第二图像331成像在感光元件平面PF2上；二维线条SA1上的点Pai与第一摄像头320的光心O1的空间连线为L1，第二图像331上二维线条SB1上的点Pb1i与第二摄像头330的光心O2的空间连线为L2，L1与L2相交于空间点P1i即为所求的Pai点对应的备选三维点之一；如果空间直线不相交，使用两条空间直线的公垂线与两条直线相交线段的中点来作为备选三维点；同样方法求得Pai点的三个备选三维点P1i、P2i和P3i(显而易见Pai点的三个备选三维点中只有一个是真实的)。
3.3、重复步骤3.1和步骤3.2，用同样方法计算得到第一图像321上的二维线条SA1上的其他点所对应的备选三维点集合。
3.4、由于备选三维点集合中的很多点并非属于真实三维点集合，需要对其进行校验筛选。在图像投影器所投的多个三维光面与摄像头 三维位置标定已知的前提下，将所述三维光面和所述第一图像321和第二图像331获得的所有备选三维点集合转换到一个坐标系下，第一图像321上的二维线条SA1上的二维点集合{Pai|1≤i≤n}对应的所有备选三维点集合为{{P1i|1≤i≤n}，{P2i|1≤i≤n}，{P3i|1≤i≤n}}，其中第二图像331中的每一条二维线条对应一个子集合，如SB1对应{P1i|1≤i≤n}。分别统计所述备选三维点集合中的每个子集合与所述三个光面之间位置关系。将子集合中的每个点到所述三个光面的距离之和作为筛选的判据：
Figure PCTCN2017086258-appb-000002
其中D(PLk,Pji)为二维线条SBj所对应的备选三维点Pji到某个光面PLk的距离。筛选得到的最小Wm＝min(Wk|k＝1,2,3)，即判定Pji∈PLm，即光面PLm上的三维点集合{Pmi|1≤i≤n}为第一图像321上的二维线条SA1对应的真实三维点集合，即光面PLm投射到物体100表面的真实三维轮廓线在第一摄像头320上成像为二维线条SA1，在第二摄像头330上的成像为二维线条SBj。
图4是本发明实施例二中真实三维点集合获得方法示意图。所述示意图包括第一摄像头410、第一图像411、第二摄像头420、第二图像421、第三摄像头430、第三图像431、图案投影器(图中未标出)。
第一摄像头410捕捉的二维图像为第一图像411，第二摄像头420捕捉的二维图像为第二图像421，第三摄像头430捕捉的二维图像为第三图像431，第一摄像头410所捕捉的第一图像411成像在感光元件平面PF1上，第二摄像头420所捕捉的第二图像421成像在感光元 件平面PF2上，第三摄像头430所捕捉的第三图像431成像在感光元件平面PF3上，O1、O2、O3分别为第一摄像头410、第二摄像头420和第三摄像头430的光心,所述摄像头的内外参已知。。
在摄像头个数为三个或更多，且所有摄像头的内外参均已知的前提下，可以利用第三个或者更多的摄像头进行数据校验。系统可以不用事先标定摄像头与图案投影器所投射的多个光面的空间位置关系，而是利用第三个摄像头来对获得的备选三维点集合中的三维点进行校验，输出真实的三维点集合。
具体方法如下：
4.1、按照所述步骤S2方法获得第三图像431上的二维线条集合为{SC1，SC2，SC3}，按照步骤3.2的方法计算获得第一图像411上的二维线条SA1上的点Pai对应的三个备选三维点P1i、P2i和P3i分别与第三摄像头430的光心O3连线，与其焦平面PF3(即第三图像431)相交于Pc1i、Pc2i和Pc3i三点，显而易见，所述三点中只有一个是真实的三维点的成像，同样的，第一图像311上的二维线条SA1上的二维点集合{Pai|1≤i≤n}对应的所有备选三维点集合为{{P1i|1≤i≤n}，{P2i|1≤i≤n}，{P3i|1≤i≤n}}，其中每个子集与第二图像321中的每一条二维线条相对应，如二维线条SB1对应子集{P1i|1≤i≤n}。
4.2、分别将备选三维点集合{{P1i|1≤i≤n}，{P2i|1≤i≤n}，{P3i|1≤i≤n}}的每个点与第三摄像头430的光心O3连线，与感光元件PF3上第三图像431的交点集合为{{Pc1i|1≤i≤n}，{Pc2i|1≤i≤n}，{Pc3i|1≤i≤n}}。分别统计所述交点集合中的每个子集合与第三图像 431上的三条二维线条{SC1，SC2，SC3}之间位置关系。将统计子集合中的每个点到某条二维线条SCk的距离的和作为筛选的判据：
Figure PCTCN2017086258-appb-000003
Figure PCTCN2017086258-appb-000004
第一图像411上的二维线条SA1上的点集{Pai|1≤i≤n}根据极线约束原理计算出极线后(N3为极点)，所述极线与二维线条SBj的交点集合所对应的备选三维点为Pji与第三摄像头430的光心O3连线相交于第三图像431的点集为{Pcji|1≤i≤n}，点集{Pcji|1≤i≤n}到二维线条SCk的距离即为D(SCk,Pcji)。筛选得到最小的Wm＝min(Wk|k＝1,2,3)，即二维线条SCm为二维线条SA1对应的真实三维点集合{Pmi|1≤i≤n}在第三图像431上的成像，也即第一图像411上的二维线条SA1和第二图像421上的二维线条SBj，以及第三图像431上的二维线条SCm为同一真实三维点集合的成像投影。
图5是本发明实施例三中真实三维点集合获得方法示意图。所述示意图包括图案投影器510、第一摄像头520、第一图像521、第二摄像头530、第二图像531。所述图案投影器510投射出三个光面PL1、PL2和PL3，第一摄像头520捕捉的二维图像为第一图像521，第二摄像头530捕捉的二维图像为第二图像531，第一摄像头520所捕捉的第一图像521成像在感光元件平面PF1上，第二摄像头530所捕捉的第二图像531成像在感光元件平面PF2上，O1、O2分别为第一摄像头520和第二摄像头530的光心，所述摄像头的内外参已知。
具体方法如下：
5.1、按照步骤S2得到第一图像521上的二维线条SA1上的二维点集合{Pai|1≤i≤n}，利用三角法原理，将所述集合与第一摄像头 520的光心O1的连线的延长线与三个光面PL1、PL2和PL3相交于{{P1i|1≤i≤n}，{P2i|1≤i≤n}，{P3i|1≤i≤n}}，即为备选三维点集合。
5.2、将上述备选三维点集合中的每个点分别与第二摄像头530的光心O2连线交第二摄像头530的感光元件平面PF2于点{{Pb1i|1≤i≤n}，{Pb2i|1≤i≤n}，{Pb3i|1≤i≤n}}，分别统计交点集合的每个子集合与第二图像531上的三条二维线条SB1、SB2和SB3之间位置关系。将子集合中的每个点到某条二维线条SBk的距离的和作为筛选的判据：
Figure PCTCN2017086258-appb-000005
第一图像521上的二维线条SA1上的点集{Pai|1≤i≤n}与光面PLj关联的备选三维点集合为{Pji|1≤i≤n}，再与第二图像531关联的点集{Pbji|1≤i≤n}到二维线条SBk的距离即为D(SBk,Pbji)。筛选得到的最小Wm＝min(Wk|k＝1,2,3)，即判断SBm为光面PLj在第二摄像头530上的成像线条，也即光面PLj上的三维点集合{Pji|1≤i≤n}为第一图像521的二维线条SA1对应的真实三维点集合，也即光面PLj投射到物体100表面的真实三维点集合在第一摄像头520上成像为二维线条SA1，在第二摄像头530上的成像为二维线条SBm。
所述三维传感器系统由于采用了多条线状图案投射，因此，其出点的效率是传统单线扫描的数倍，显著的提升了扫描的速度。
本发明实施例还提供了一种三维数据获取方法，所述方法包括以下步骤：
步骤101、图案投影器投射出至少两条线状图案；
优选的，至少两条线状图案可以为一个所述图案投影器投射，或多个所述图案投影器同时投射；所述线状图案为直线或曲线线条。所述图案投影器包括一线状激光器和一DOE分束元件，所述线状激光器通过所述DOE分束元件分出多条激光线段。
优选的，所述图案投影器也可以为一投影仪，所述投影仪直接投出所述至少两条线状图案。
步骤102、至少两摄像头同步捕捉二维图像；
步骤103、提取所述二维图像上被扫描物体表面线状投影的二维线条集合；
优选的，所述提取所述二维图像上被扫描物体表面线状投影的二维线条集合，具体包括：根据所述二维图像所对应摄像头的内参对所述二维图像进行畸变矫正，并根据像素灰度差异提取矫正图像中线条轮廓的连通区域，再根据所述连通区域内的灰度重心计算获得亚像素级的高光中心二维线条集合。
步骤104、将所述二维线条集合生成备选三维点集合；
步骤105、从所述备选三维点集合中筛选出正确匹配物体表面投影轮廓线的真实三维点集合。
具体的，步骤104和步骤105可以通过以下3种方法实现：
第一种方法为：从至少两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述摄像头之间的空间位置关系，根据三角法原理和极线约束原理计算得出所述备选三维点集合；根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来 判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合。
优选的，所述备选三维点集合包括若干子集合，以所述子集合到所述三维光面的距离作为依据，筛选出距离最小的子集合即为真实的三维点集合。
第二种方法为：从两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述两幅同步二维图像对应的两个摄像头的空间位置关系，并根据三角法原理和极线约束原理计算得出所述备选三维点集合；利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合。
优选的，所述备选三维点集合包括若干子集合，所述子集合与所述第三个摄像头的光心连线与所述第三个摄像头所拍摄的二维图像存在交点集合，以所述交点集合到所述第三个摄像头所拍摄的二维图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
第三种方法为：从任一摄像头所捕捉图像的所述二维线条集合中提取二维点数据，利用所述图案投影器所投射的多个空间光面与该摄像头的空间位置关系，根据三角法原理得出所述备选三维点集合；将所述备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合。
优选的，所述备选三维点集合包括若干子集合，所述子集合与所述另外至少一个摄像头的光心连线与所述另外至少一个摄像头所拍 摄的图像存在交点集合，以所述交点集合到所述另外至少一个摄像头所拍摄的图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
需要说明的是，本实施例中的三维数据获取方法与前述实施例中的三维传感器系统是基于同一发明构思下的两个方面，在前面已经对方法实施过程作了详细的描述，所以本领域技术人员可根据前述描述清楚地了解本实施中的系统的结构及实施过程，为了说明书的简洁，在此就不再赘述。
对于本领域的技术人员来说，可根据以上描述的技术方案以及构思，做出其它各种相应的改变以及变形，而所有的这些改变以及变形都应该属于本发明权利要求的保护范围之内。

Claims (20)

  1. 一种三维传感器系统，其特征在于，包括至少一图案投影器、至少两摄像头、一二维图像特征提取器、一三维点云生成器、一三维点云校验器；
    所述图案投影器，用于同时投射出至少两条线状图案；
    所述至少两摄像头，用于同步捕捉被扫描物体的二维图像；
    所述二维图像特征提取器，用于提取所述二维图像上被扫描物体表面所述至少两条线状图案的二维线条集合；
    所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合；
    所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合。
  2. 如权利要求1所述的三维传感器系统，其特征在于，所述二维图像特征提取器，用于提取所述二维图像上被扫描物体表面所述至少两条线状图案的二维线条集合，具体包括：所述二维图像特征提取器根据所述二维图像所对应摄像头的内参对所述二维图像进行畸变矫正，并根据像素灰度差异提取矫正图像中线条轮廓的连通区域，再根据所述连通区域内的灰度重心计算获得亚像素级的高光中心二维线条集合。
  3. 如权利要求1或2所述的三维传感器系统，其特征在于，所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合，具体包括：所述三维点云生成器从至少两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述摄像头之间的空间位置关系，根 据三角法原理和极线约束原理计算得出所述备选三维点集合；
    所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：所述三维点云校验器根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合。
  4. 如权利要求3所述的三维传感器系统，其特征在于，所述三维点云校验器根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所述三维点云校验器以所述子集合到所述三维光面的距离作为依据，筛选出距离最小的子集合即为真实的三维点集合。
  5. 如权利要求1或2所述的三维传感器系统，其特征在于，所述摄像头数量为至少三个；
    所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合，具体包括：所述三维点云生成器从两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述两幅同步二维图像对应的两个摄像头的空间位置关系，并根据三角法原理和极线约束原理计算得出所述备选三维点集合；
    所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：所述三维点云校验器利用第三个或者更多的摄像头所拍摄的二维图像对所述 备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合。
  6. 如权利要求5所述的三维传感器系统，其特征在于，所述三维点云校验器利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所述子集合与所述第三个摄像头的光心连线与所述第三个摄像头所拍摄的二维图像存在交点集合，所述三维点云校验器以所述交点集合到所述第三个摄像头所拍摄的二维图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
  7. 如权利要求1或2所述的三维传感器系统，其特征在于，所述三维点云生成器，用于将所述二维线条集合生成备选三维点集合，具体包括：所述三维点云生成器从任一摄像头所捕捉图像的所述二维线条集合中提取二维点数据，利用所述图案投影器所投射的多个空间光面与该摄像头的空间位置关系，根据三角法原理得出所述备选三维点集合；
    所述三维点云校验器，用于从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：所述三维点云校验器将所述备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合。
  8. 如权利要求7所述的三维传感器系统，其特征在于，所述三维点云校验器将所述备选三维点集合与另外至少一个摄像头所拍摄的图像进行校验，并进行筛选得到真实的三维点集合，具体包括：所述 备选三维点集合包括若干子集合，所述子集合与所述另外至少一个摄像头的光心连线与所述另外至少一个摄像头所拍摄的图像存在交点集合，所述三维点云校验器以所述交点集合到所述另外至少一个摄像头所拍摄的图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
  9. 如权利要求1所述的三维传感器系统，其特征在于，所述线状图案为一个所述图案投影器投射，或多个所述图案投影器同时投射；所述线状图案为直线或曲线线条。
  10. 如权利要求1所述的三维传感器系统，其特征在于，所述图案投影器包括一线状激光器和一DOE分束元件，所述线状激光器通过所述DOE分束元件分出多条激光线段。
  11. 如权利要求1所述的三维传感器系统，其特征在于，所述图案投影器包括一投影仪，所述投影仪直接投出所述至少两条线状图案。
  12. 如权利要求1所述的三维传感器系统，其特征在于，所述三维传感器系统包括一同步触发器，所述同步触发器用于触发所述摄像头和所述图案投影器进行同步工作。
  13. 一种三维数据获取方法，其特征在于，包括以下步骤：
    图案投影器投射出至少两条线状图案；
    至少两摄像头同步捕捉二维图像；
    提取所述二维图像上被扫描物体表面至少两条线状图案的二维线条集合；
    将所述二维线条集合生成备选三维点集合；
    从所述备选三维点集合中筛选出正确匹配物体表面投影轮廓线的真实三维点集合。
  14. 如权利要求13所述的三维数据获取方法，其特征在于，所述提取所述二维图像上被扫描物体表面至少两条线状图案的二维线条集合，具体包括：根据所述二维图像所对应摄像头的内参对所述二维图像进行畸变矫正，并根据像素灰度差异提取矫正图像中线条轮廓的连通区域，再根据所述连通区域内的灰度重心计算获得亚像素级的高光中心二维线条集合。
  15. 如权利要求13或14所述的三维数据获取方法，其特征在于，将所述二维线条集合生成备选三维点集合，具体包括：从至少两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述摄像头之间的空间位置关系，根据三角法原理和极线约束原理计算得出所述备选三维点集合；
    从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：根据所述备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合。
  16. 如权利要求15所述的三维数据获取方法，其特征在于，所述根据备选三维点集合中的点是否处于所述图案投影器所投射的某个三维光面来判断该点是否属于真实的三维点集合，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，以所述子集合到所述三维光面的距离作为依据，筛选出距离最小的子 集合即为真实的三维点集合。
  17. 如权利要求13或14所述的三维数据获取方法，其特征在于，所述摄像头数量为至少三；
    将所述二维线条集合生成备选三维点集合，具体包括：从两幅同步二维图像的二维线条集合中分别提取二维点数据，利用所述两幅同步二维图像对应的两个摄像头的空间位置关系，并根据三角法原理和极线约束原理计算得出所述备选三维点集合；
    从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合。
  18. 如权利要求17所述的三维数据获取方法，其特征在于，所述利用第三个或者更多的摄像头所拍摄的二维图像对所述备选三维点集合进行数据校验，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所述子集合与所述第三个摄像头的光心连线与所述第三个摄像头所拍摄的二维图像存在交点集合，以所述交点集合到所述第三个摄像头所拍摄的二维图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
  19. 如权利要求13或14所述的三维数据获取方法，其特征在于，将所述二维线条集合生成备选三维点集合，具体包括：从任一摄像头所捕捉图像的所述二维线条集合中提取二维点数据，利用所述图案投 影器所投射的多个空间光面与该摄像头的空间位置关系，根据三角法原理得出所述备选三维点集合；
    从所述备选三维点集合中筛选出正确匹配物体表面的投影轮廓线的真实三维点集合，具体包括：将所述备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合。
  20. 如权利要求19所述的三维数据获取方法，其特征在于，所述将备选三维点集合与另外至少一个摄像头的图像进行校验，并进行筛选得到真实的三维点集合，具体包括：所述备选三维点集合包括若干子集合，所述子集合与所述另外至少一个摄像头的光心连线与所述另外至少一个摄像头所拍摄的图像存在交点集合，以所述交点集合到所述另外至少一个摄像头所拍摄的图像上二维线条的距离作为依据，筛选出距离最小值所对应的子集合即为真实的三维点集合。
PCT/CN2017/086258 2016-09-14 2017-05-27 三维传感器系统及三维数据获取方法 Ceased WO2018049843A1 (zh)

Priority Applications (5)

Application Number Priority Date Filing Date Title
EP21177443.5A EP3907702B1 (en) 2016-09-14 2017-05-27 Three-dimensional sensor system and three-dimensional data acquisition method
US15/573,487 US10309770B2 (en) 2016-09-14 2017-05-27 Three-dimensional sensor system and three-dimensional data acquisition method
KR1020197010317A KR102096806B1 (ko) 2016-09-14 2017-05-27 3차원 센서 시스템 및 3차원 데이터 획득방법
EP17850048.4A EP3392831B1 (en) 2016-09-14 2017-05-27 Three-dimensional sensor system and three-dimensional data acquisition method
US16/428,007 US11060853B2 (en) 2016-09-14 2019-05-31 Three-dimensional sensor system and three-dimensional data acquisition method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610824489.3 2016-09-14
CN201610824489.3A CN106403845B (zh) 2016-09-14 2016-09-14 三维传感器系统及三维数据获取方法

Related Child Applications (2)

Application Number Title Priority Date Filing Date
US15/573,487 A-371-Of-International US10309770B2 (en) 2016-09-14 2017-05-27 Three-dimensional sensor system and three-dimensional data acquisition method
US16/428,007 Continuation-In-Part US11060853B2 (en) 2016-09-14 2019-05-31 Three-dimensional sensor system and three-dimensional data acquisition method

Publications (1)

Publication Number Publication Date
WO2018049843A1 true WO2018049843A1 (zh) 2018-03-22

Family

ID=57998097

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/086258 Ceased WO2018049843A1 (zh) 2016-09-14 2017-05-27 三维传感器系统及三维数据获取方法

Country Status (5)

Country Link
US (1) US10309770B2 (zh)
EP (2) EP3907702B1 (zh)
KR (1) KR102096806B1 (zh)
CN (1) CN106403845B (zh)
WO (1) WO2018049843A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110189326A (zh) * 2019-04-09 2019-08-30 广东元一科技实业有限公司 一种基于二维图像的人体数据测量系统及其测量方法
CN110415331A (zh) * 2019-06-14 2019-11-05 中国电建集团昆明勘测设计研究院有限公司 一种基于点云数据的轮廓或孔洞快速检测判别和孔洞修补方法
CN113436136A (zh) * 2020-03-06 2021-09-24 苹果公司 以对象为中心的扫描
CN115995010A (zh) * 2023-03-22 2023-04-21 华南农业大学 一种基于植株群体点云的株高提取方法及系统
CN116563666A (zh) * 2023-05-09 2023-08-08 京东科技信息技术有限公司 三维点云处理模型和语义分割模型的训练方法和相关设备
CN117333649A (zh) * 2023-10-25 2024-01-02 天津大学 一种动态扰动下高频线扫描稠密点云的优化方法

Families Citing this family (39)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10643343B2 (en) 2014-02-05 2020-05-05 Creaform Inc. Structured light matching of a set of curves from three cameras
KR102424135B1 (ko) 2014-02-05 2022-07-25 크레아폼 인크. 2개의 카메라로부터의 곡선의 세트의 구조형 광 매칭
CN106403845B (zh) 2016-09-14 2017-10-03 杭州思看科技有限公司 三维传感器系统及三维数据获取方法
US10996713B2 (en) * 2017-08-07 2021-05-04 Apple Inc. Portable electronic device
JP2019047311A (ja) * 2017-09-01 2019-03-22 セイコーエプソン株式会社 画像投写システム及びその制御方法
JP6926948B2 (ja) * 2017-10-27 2021-08-25 セイコーエプソン株式会社 プロジェクター、画像投写システム、及びプロジェクターの制御方法
DE102018115334B3 (de) * 2018-06-26 2019-05-09 Dr. Ing. H.C. F. Porsche Aktiengesellschaft Verfahren zum Kalibrieren einer elektromagnetische Strahlung abstrahlenden Vorrichtung mittels einer Sensoreinheit
CN109141289B (zh) * 2018-08-01 2020-12-29 先临三维科技股份有限公司 三维扫描方法和系统
WO2020061792A1 (en) * 2018-09-26 2020-04-02 Intel Corporation Real-time multi-view detection of objects in multi-camera environments
CN109341591A (zh) * 2018-11-12 2019-02-15 杭州思看科技有限公司 一种基于手持式三维扫描仪的边缘检测方法及系统
JP7139953B2 (ja) * 2019-01-08 2022-09-21 オムロン株式会社 三次元形状測定装置、三次元形状測定方法及びプログラム
CN111623725B (zh) * 2019-02-27 2022-04-29 杭州思看科技有限公司 一种跟踪式三维扫描系统
CN109903327B (zh) * 2019-03-04 2021-08-31 西安电子科技大学 一种稀疏点云的目标物尺寸测量方法
WO2020208686A1 (ja) * 2019-04-08 2020-10-15 日本電気株式会社 カメラ校正装置、カメラ校正方法、及びプログラムが格納された非一時的なコンピュータ可読媒体
EP3970362A4 (en) * 2019-05-12 2023-06-21 Magik Eye Inc. MAPPING OF THREE-DIMENSIONAL DEPTH MAP DATA TO TWO-DIMENSIONAL IMAGES
CN110186396B (zh) * 2019-06-05 2021-04-30 中南大学 一种获取tbm隧洞形貌三维点云数据的装置及方法
KR102161452B1 (ko) * 2019-07-24 2020-10-05 (주)칼리온 스캐너 움직임에 따른 모션 유효성 검출 장치 및 그 방법
WO2021025879A2 (en) * 2019-08-07 2021-02-11 Size Stream Llc Mobile 3d crowd scanning methods and apparatus
US20210156881A1 (en) * 2019-11-26 2021-05-27 Faro Technologies, Inc. Dynamic machine vision sensor (dmvs) that performs integrated 3d tracking
DE102020201536A1 (de) 2020-02-07 2021-08-12 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung eingetragener Verein Verfahren und Vorrichtung zum berührungslosen Vermessen einer Objektoberfläche
CN111340889B (zh) * 2020-02-19 2023-04-07 厦门大学 基于车载激光扫描自动获取匹配图像块与点云球的方法
CN111678459B (zh) * 2020-06-09 2021-10-08 杭州思看科技有限公司 三维扫描方法、三维扫描系统和计算机可读存储介质
CN111739359A (zh) * 2020-06-30 2020-10-02 上海乂学教育科技有限公司 增强现实的课件生成系统
KR20220026238A (ko) * 2020-08-25 2022-03-04 주식회사 메디트 복수의 이미지 센서를 구비하는 3차원 스캐너
CN112525106B (zh) * 2020-10-23 2022-08-26 清华大学 基于三相机协同激光的3d探测方法及装置
CN112747671B (zh) * 2020-12-10 2022-12-09 杭州先临天远三维检测技术有限公司 三维检测系统和三维检测方法
US12487073B2 (en) 2020-12-23 2025-12-02 Faro Technologies, Inc. Speckle reduction methods in line scanners
US12483625B2 (en) 2020-12-23 2025-11-25 Faro Technologies, Inc. Handheld scanner that transfers data and power over ethernet
US20220316869A1 (en) * 2020-12-23 2022-10-06 Faro Technologies, Inc. Line scanner having integrated processing capability
US11908162B2 (en) 2020-12-23 2024-02-20 Faro Technologies, Inc. Line scanner having target-tracking and geometry-tracking modes
DE102021108232A1 (de) 2021-03-31 2022-10-06 Carl Zeiss Industrielle Messtechnik Gmbh Messgerät zur 3D-Erfassung einer Oberfläche
CN114898038B (zh) * 2022-04-28 2025-12-23 杭州海康机器人股份有限公司 一种图像重建方法、装置及设备
CN115615352B (zh) * 2022-09-28 2023-07-11 山东交通学院 激光扫描器的标定方法、装置以及存储介质
CN115598614A (zh) * 2022-11-28 2023-01-13 南京隼眼电子科技有限公司（Cn） 三维点云目标检测方法、装置及存储介质
WO2025137175A1 (en) * 2023-12-19 2025-06-26 Hover Inc. Methods, storage media, and systems for generating a three-dimensional line segment
CN117689965B (zh) * 2024-02-04 2025-01-28 山东黄金矿业股份有限公司新城金矿 基于ransac算法的激光点云岩体结构快速识别方法
CN118781507B (zh) * 2024-09-12 2024-11-22 成都经开地理信息勘测设计院有限公司 基于车载点云和航片的道路信息识别方法及系统
CN119289864B (zh) * 2024-12-10 2025-04-01 武汉中观自动化科技有限公司 一种三维点云采集方法和结构光扫描设备
CN119785133B (zh) * 2025-03-13 2025-06-27 南京航空航天大学 一种多模态视觉数据融合的曲面天线关键特征重构方法

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101908230A (zh) * 2010-07-23 2010-12-08 东南大学 一种基于区域深度边缘检测和双目立体匹配的三维重建方法
US20100328308A1 (en) * 2008-07-10 2010-12-30 C-True Ltd. Three Dimensional Mesh Modeling
CN103930924A (zh) * 2011-08-09 2014-07-16 依莫菲克斯公司 图像处理方法
CN104408762A (zh) * 2014-10-30 2015-03-11 福州大学 利用单目和二维平台获取物体图像信息及三维模型的方法
CN104463826A (zh) * 2013-09-18 2015-03-25 镇江福人网络科技有限公司 一种新的点云并行Softassign配准算法
CN104778680A (zh) * 2014-01-15 2015-07-15 波音公司 检查目标的系统和方法
CN105261060A (zh) * 2015-07-23 2016-01-20 东华大学 基于点云压缩和惯性导航的移动场景实时三维重构方法
CN106403845A (zh) * 2016-09-14 2017-02-15 杭州思看科技有限公司 三维传感器系统及三维数据获取方法

Family Cites Families (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5589942A (en) * 1990-04-05 1996-12-31 Intelligent Automation Systems Real time three dimensional sensing system
WO2000057129A1 (en) * 1999-03-19 2000-09-28 Matsushita Electric Works, Ltd. Three-dimensional object recognition method and pin picking system using the method
JP2005515487A (ja) * 2002-01-04 2005-05-26 ニューローケイ・エルエルシー 再帰反射スクリーンを使用した３次元画像投影
DE10219054B4 (de) * 2002-04-24 2004-08-26 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Verfahren und Vorrichtung zur Bestimmung der räumlichen Koordinaten eines Gegenstandes
US6944564B2 (en) * 2002-05-08 2005-09-13 Metris N.V. Method for the automatic calibration-only, or calibration and qualification simultaneously of a non-contact probe
US20100014099A1 (en) * 2004-12-16 2010-01-21 Werth Messtechnik Gmbh Coordinate measuring device and method for measuring with a coordinate measuring device
WO2010015086A1 (en) * 2008-08-06 2010-02-11 Creaform Inc. System for adaptive three-dimensional scanning of surface characteristics
TWI413923B (zh) * 2009-06-30 2013-11-01 E Pin Optical Industry Co Ltd 微機電掃描之座標偵測方法及其觸控螢幕
EP2400261A1 (de) * 2010-06-21 2011-12-28 Leica Geosystems AG Optisches Messverfahren und Messsystem zum Bestimmen von 3D-Koordinaten auf einer Messobjekt-Oberfläche
US9686532B2 (en) * 2011-04-15 2017-06-20 Faro Technologies, Inc. System and method of acquiring three-dimensional coordinates using multiple coordinate measurement devices
JP2013210254A (ja) * 2012-03-30 2013-10-10 Canon Inc 三次元計測装置、三次元計測方法及び三次元計測プログラム
US20150015701A1 (en) * 2013-07-10 2015-01-15 Faro Technologies, Inc. Triangulation scanner having motorized elements
CN104517318A (zh) * 2013-09-27 2015-04-15 鸿富锦精密工业（深圳）有限公司 三维量测模拟取点系统及方法
US9619691B2 (en) * 2014-03-07 2017-04-11 University Of Southern California Multi-view 3D object recognition from a point cloud and change detection
CN103983207A (zh) * 2014-05-30 2014-08-13 深圳先进技术研究院 一种三维扫描内窥镜和三维扫描方法
CN104165600B (zh) * 2014-07-03 2016-10-05 杭州鼎热科技有限公司 一种无线手持3d激光扫描系统
US20160003612A1 (en) * 2014-07-07 2016-01-07 VirtualU Rapid and accurate three dimensional scanner
US9674504B1 (en) * 2015-12-22 2017-06-06 Aquifi, Inc. Depth perceptive trinocular camera system

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100328308A1 (en) * 2008-07-10 2010-12-30 C-True Ltd. Three Dimensional Mesh Modeling
CN101908230A (zh) * 2010-07-23 2010-12-08 东南大学 一种基于区域深度边缘检测和双目立体匹配的三维重建方法
CN103930924A (zh) * 2011-08-09 2014-07-16 依莫菲克斯公司 图像处理方法
CN104463826A (zh) * 2013-09-18 2015-03-25 镇江福人网络科技有限公司 一种新的点云并行Softassign配准算法
CN104778680A (zh) * 2014-01-15 2015-07-15 波音公司 检查目标的系统和方法
CN104408762A (zh) * 2014-10-30 2015-03-11 福州大学 利用单目和二维平台获取物体图像信息及三维模型的方法
CN105261060A (zh) * 2015-07-23 2016-01-20 东华大学 基于点云压缩和惯性导航的移动场景实时三维重构方法
CN106403845A (zh) * 2016-09-14 2017-02-15 杭州思看科技有限公司 三维传感器系统及三维数据获取方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3392831A4 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110189326A (zh) * 2019-04-09 2019-08-30 广东元一科技实业有限公司 一种基于二维图像的人体数据测量系统及其测量方法
CN110415331A (zh) * 2019-06-14 2019-11-05 中国电建集团昆明勘测设计研究院有限公司 一种基于点云数据的轮廓或孔洞快速检测判别和孔洞修补方法
CN110415331B (zh) * 2019-06-14 2022-08-19 中国电建集团昆明勘测设计研究院有限公司 一种基于点云数据的轮廓或孔洞快速检测判别和孔洞修补方法
CN113436136A (zh) * 2020-03-06 2021-09-24 苹果公司 以对象为中心的扫描
CN115995010A (zh) * 2023-03-22 2023-04-21 华南农业大学 一种基于植株群体点云的株高提取方法及系统
CN115995010B (zh) * 2023-03-22 2023-06-16 华南农业大学 一种基于植株群体点云的株高提取方法及系统
CN116563666A (zh) * 2023-05-09 2023-08-08 京东科技信息技术有限公司 三维点云处理模型和语义分割模型的训练方法和相关设备
CN117333649A (zh) * 2023-10-25 2024-01-02 天津大学 一种动态扰动下高频线扫描稠密点云的优化方法
CN117333649B (zh) * 2023-10-25 2024-06-04 天津大学 一种动态扰动下高频线扫描稠密点云的优化方法

Also Published As

Publication number Publication date
KR102096806B1 (ko) 2020-04-03
US20180356213A1 (en) 2018-12-13
EP3907702A1 (en) 2021-11-10
EP3392831A1 (en) 2018-10-24
EP3907702B1 (en) 2024-03-20
CN106403845A (zh) 2017-02-15
EP3392831A4 (en) 2020-01-01
EP3392831B1 (en) 2021-07-14
KR20190041537A (ko) 2019-04-22
US10309770B2 (en) 2019-06-04
CN106403845B (zh) 2017-10-03

Similar Documents

Publication Publication Date Title
WO2018049843A1 (zh) 三维传感器系统及三维数据获取方法
CN105550670B (zh) 一种目标物体动态跟踪与测量定位方法
CN106683070B (zh) 基于深度相机的身高测量方法及装置
EP3650807B1 (en) Handheld large-scale three-dimensional measurement scanner system simultaneously having photography measurement and three-dimensional scanning functions
US11060853B2 (en) Three-dimensional sensor system and three-dimensional data acquisition method
US8218858B2 (en) Enhanced object reconstruction
JP5132832B1 (ja) 計測装置および情報処理装置
KR102424135B1 (ko) 2개의 카메라로부터의 곡선의 세트의 구조형 광 매칭
CN104567727B (zh) 利用立体靶标对线结构光轮廓传感器的全局统一校准方法
US20150003726A1 (en) Semi-supervised method for training multiple pattern recognition and registration tool models
KR101759798B1 (ko) 실내 2d 평면도의 생성 방법, 장치 및 시스템
CN106033614B (zh) 一种强视差下的移动相机运动目标检测方法
JP7489253B2 (ja) デプスマップ生成装置及びそのプログラム、並びに、デプスマップ生成システム
CN108010125A (zh) 基于线结构光和图像信息的真实尺度三维重建系统及方法
CN115601438B (zh) 外参标定方法、装置以及自主移动设备
Furferi et al. A RGB-D based instant body-scanning solution for compact box installation
JP2016156702A (ja) 撮像装置および撮像方法
JP6121746B2 (ja) 撮影位置決定装置、プログラム、および方法
CN116246038B (zh) 一种多视影像三维线段重建方法、系统、电子设备及介质
Ortiz-Coder et al. Accurate 3d reconstruction using a videogrammetric device for heritage scenarios
Zhang et al. Combining kinect and pnp for camera pose estimation
JP6061631B2 (ja) 計測装置、情報処理装置、計測方法、情報処理方法、および、プログラム
CN207637202U (zh) 基于线结构光和图像信息的真实尺度三维重建系统
Nebel et al. Stroboscopic stereo rangefinder
Kawasaki et al. Registration and entire shape acquisition for grid based active one-shot scanning techniques

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 2017850048

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2017850048

Country of ref document: EP

Effective date: 20180718

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 20197010317

Country of ref document: KR

Kind code of ref document: A