WO2020001630A1 - 一种三值图像获取方法、装置及车辆 - Google Patents

一种三值图像获取方法、装置及车辆 Download PDF

Info

Publication number
WO2020001630A1
WO2020001630A1 PCT/CN2019/093770 CN2019093770W WO2020001630A1 WO 2020001630 A1 WO2020001630 A1 WO 2020001630A1 CN 2019093770 W CN2019093770 W CN 2019093770W WO 2020001630 A1 WO2020001630 A1 WO 2020001630A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
image data
ternary
differential
identified
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/CN2019/093770
Other languages
English (en)
French (fr)
Inventor
宫原俊二
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.)
Great Wall Motor Co Ltd
Original Assignee
Great Wall Motor 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 Great Wall Motor Co Ltd filed Critical Great Wall Motor Co Ltd
Priority to EP19824956.7A priority Critical patent/EP3800614A4/en
Priority to US17/256,844 priority patent/US11861844B2/en
Publication of WO2020001630A1 publication Critical patent/WO2020001630A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20004—Adaptive image processing
    • G06T2207/20012—Locally adaptive
    • 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
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/07—Target detection

Definitions

  • the invention relates to the technical field of automatic driving, and includes a method, a device and a vehicle for acquiring a three-valued image.
  • ADAS Advanced Driver Assistance Systems
  • ADAS Advanced Driver Assistance Systems
  • the system performs calculation and analysis, so that the driver can be aware of possible dangers in advance, effectively increasing the comfort and safety of car driving, especially in the field of autonomous driving, various sensors in ADAS
  • vision cameras have the function of obtaining two-dimensional images like humans, so they can be used as the main sensor in ADAS in autonomous vehicles.
  • the vision camera uses image processing functions to capture road images while the vehicle is driving, and assists the driver to accurately identify the road in the automatic driving mode, which is an important subject in the field of automatic driving.
  • the most important vision camera image processing technology is edge extraction, and by acquiring the three-valued image of the detection object, it is more effective to detect the edge information in the image, and the edge extraction to capture the object and environment information in the image.
  • edge extraction is the key technology of vision cameras, there is no effective and systematic method to obtain the three-valued image of the original image with better effect, so as to extract the edge image.
  • a three-valued image acquisition method comprising: acquiring first differential image data of the image data in a vehicle;
  • a ternary image is obtained by applying the ternary threshold on the differential image data.
  • the step of obtaining the first differential image data of the image data includes:
  • the grayscale image is processed by the Sauber Sobel algorithm to obtain first differential image data of the image data.
  • the step of obtaining a ternary coefficient by performing preset processing on an image in an image database corresponding to the object to be identified includes:
  • a ratio of a separation noise and a signal in the second differential image data is obtained according to the difference value distribution and the difference value cumulative distribution, and is determined as a ternary coefficient.
  • the step of obtaining a ternary coefficient by performing preset processing on an image in an image database corresponding to the object to be identified includes:
  • a ternary coefficient is obtained according to the cumulative distribution probability of the difference value.
  • the image database includes historical image data obtained by the object to be identified within a preset time period.
  • the three-value image acquisition method has the following advantages: by acquiring image data of an object to be identified during the driving of a vehicle; acquiring first differential image data of the image data; The image in the image database of the object to be identified is subjected to preset processing to obtain a ternary coefficient; a ternary threshold is obtained according to a product of the ternary coefficient and a maximum pixel value in the first differential image data. Obtaining a ternary image by applying the ternary threshold on the differential image data.
  • the system has the beneficial effect of systematically obtaining the ternary image of the original image.
  • Another object of the present invention is to provide a three-valued image acquisition device, which includes a recognition object image acquisition module configured to acquire image data of an object to be identified during a driving process of the vehicle;
  • a differential image data acquisition module configured to acquire first differential image data of the image data
  • a ternary threshold acquisition module configured to obtain a ternary threshold according to a product of the ternary coefficient and a maximum pixel value in the first difference image data
  • a three-valued image acquisition module is configured to obtain a three-valued image by applying the three-valued threshold on the differential image data.
  • differential image data acquisition module includes:
  • a grayscale image acquisition submodule configured to perform grayscale processing on the image data to obtain a grayscale image of the image data of the object to be identified
  • a differential image data acquisition submodule is configured to process the grayscale image by using a Sauber Sobel algorithm to obtain first differential image data of the image data.
  • the ternary coefficient acquisition module includes:
  • a second differential image data acquisition submodule configured to perform differential processing on images in an image database corresponding to the object to be identified to obtain second differential image data
  • a difference distribution acquisition submodule configured to obtain a difference value distribution and a difference value cumulative distribution of the second difference image data by normalizing the second difference image data
  • the ternization coefficient acquisition sub-module is configured to obtain a ratio of a separation noise and a signal in the second difference image data according to the difference value distribution and the difference value cumulative distribution, and determine the ternization coefficient.
  • the ternary coefficient acquisition module includes:
  • a grayscale image acquisition submodule configured to perform grayscale processing on an image in an image database corresponding to the object to be identified to obtain a grayscale image of the image data
  • a test sub-module configured to perform a three-value processing on the grayscale image to obtain a plurality of test three-valued images including pixel data of the object to be identified;
  • a pixel number acquisition sub-module configured to determine the number of pixels in each of the test three-valued images according to a preset number of pixels and a preset pixel appearance probability
  • a distribution value acquisition submodule configured to acquire a cumulative distribution probability of difference values of the pixel data according to the number of pixels in the test ternary image
  • the ternization coefficient acquisition submodule is configured to acquire a ternization coefficient according to the cumulative distribution probability of the difference value.
  • the image database includes historical image data obtained by the object to be identified within a preset time period.
  • the three-valued image acquisition device has the same advantages as the foregoing three-valued image acquisition method over the prior art, and is not repeated here.
  • Another object of the present invention is to provide a vehicle, which includes the three-valued image acquisition method and device described above.
  • Yet another object of the present invention is to provide a computer program including computer-readable code that, when the computer-readable code runs on a computing processing device, causes the computing processing device to execute a method for acquiring a three-valued image.
  • Another object of the present invention is to provide a computer-readable medium in which the computer program described above is stored.
  • FIG. 1 is a flowchart of a method for acquiring a three-valued image according to the first embodiment of the present invention
  • FIG. 2 is a schematic diagram of an image processing result in an embodiment of the present invention.
  • FIG. 3 is a schematic diagram of an image processing result in an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of an image processing principle in an embodiment of the present invention.
  • FIG. 5 is a schematic diagram of an image processing result in an embodiment of the present invention.
  • FIG. 6 is a schematic diagram of an image processing principle in an embodiment of the present invention.
  • FIG. 7 is a flowchart of a method for acquiring a three-value image according to Embodiment 2 of the present invention.
  • FIG. 8 is a schematic diagram of an image processing principle in an embodiment of the present invention.
  • FIG. 9 is a schematic diagram of an image processing principle in an embodiment of the present invention.
  • FIG. 10 is a schematic flowchart of an image processing process according to an embodiment of the present invention.
  • FIG. 11 is a structural block diagram of a three-valued image acquisition device according to Embodiment 3 of the present invention.
  • FIG. 12 is a structural block diagram of a three-valued image acquisition device according to Embodiment 3 of the present invention.
  • FIG. 13 is a schematic structural diagram of a computer processing device according to an embodiment of the present invention.
  • FIG. 14 is a schematic structural diagram of a storage unit according to an embodiment of the present invention.
  • a flowchart of a method for acquiring a three-level image may specifically include the following steps:
  • step 101 image data of an object to be identified is acquired during the running of the vehicle.
  • a vision system of a vehicle device is used to collect an object to be identified during vehicle running.
  • Figure 2 can be a global image obtained from the camera of the device in the vision system, where the image range marked in the white box is the current vehicle's main recognition object as the driving road, in order to improve the efficiency of image processing in the later stage, as shown in Figure 3,
  • the post-image processing mainly focuses on the area image in the white box.
  • Step 102 Obtain first differential image data of the image data.
  • image processing may be performed on the image through a Sobel filter to obtain first differential image data.
  • a Sobel filter As shown in FIG. 3, after the image to be processed is resampled and smoothed, the image is processed by a Sobel filter to obtain differential image data.
  • Raster Scan scans every pixel in the image in the horizontal direction, and transforms XXX (m, n) on the original image into differential image differential image data ddd (m, n ), You can refer to the following code:
  • Step 103 Perform preset processing on an image in an image database corresponding to the object to be identified to obtain a ternary coefficient.
  • the differential image is subjected to three-level processing:
  • the image data of the object to be identified is repeatedly subjected to three-level binarization processing to obtain an optimal recognition result experience, and a forward boundary pixel or The range of cumulative distribution probability of negative boundary pixels.
  • formula (2) the differential value distribution map of the positive boundary pixels of the above differential image data, and the cumulative distribution probability of the positive boundary pixels or negative boundary pixels are obtained, and the cumulative distribution probability corresponding to the above experience is inferred inversely.
  • the ternary coefficient of the range is inversely.
  • a plurality of three-valued images obtained from negative boundary pixels in multiple difference images. According to the test and observation of the researchers, these four images respectively meet the preset conditions, which reflects the noise.
  • the features of fewer and clear boundaries, and the cumulative distribution probability range of the forward boundary pixels corresponding to the above four images are between 97% and 99%, and the probability distribution range is the ternary coefficient.
  • the difference value distribution map of the positive boundary pixels corresponding to the difference map shown in FIG. 3 (the upper part of FIG. 6), and the cumulative distribution probability of the positive boundary pixels to the negative boundary pixels ( (Bottom half of FIG. 6), according to the previously obtained preset cumulative distribution probability (threshold) is between 97% -99%, the best is in the difference value distribution chart shown in the upper half of FIG.
  • the cumulative distribution probability is 97% -99% and the corresponding ternization coefficient (ratio) is 0.1-0.4.
  • Step 104 Obtain a ternary threshold value according to a product of the ternary coefficient and a maximum pixel value in the first difference image data.
  • the obtained trinization coefficient is multiplied by the maximum pixel value in the obtained first difference image data, and the obtained trinization threshold is obtained.
  • the ternary coefficient is 0.3
  • FIG. 6 (a) according to the histogram distribution of the maximum pixel value coefficient, 99% of the corresponding ternary threshold can be obtained.
  • step 105 a ternary image is obtained by applying the ternary threshold on the differential image data.
  • the obtained differential image data is further subjected to the three-valued processing by using the obtained three-valued coefficients to obtain the three-valued image of the differential image data.
  • the value of each pixel in the obtained three-level image ttt (m, n) is selected from the set [-1,0,1]. Therefore, the pixels in the differential image ddd (m, n) that are larger than the preset threshold Threshold are assigned a value of 1, where the value of Threshold is ratio * max (
  • the ratios obtained according to the ternary coefficients are respectively 0.1, 0.2, 0.3, and 0.4 to obtain a better ternary. image.
  • the image data of the object to be identified is acquired during the running of the vehicle; the first differential image data of the image data is acquired; and the image in the image database corresponding to the object to be identified is pre-predicted.
  • FIG. 7 is a flowchart of a method for acquiring a three-value image according to an embodiment of the present invention, which specifically includes the following steps:
  • Step 201 Obtain image data of an object to be identified during the running of the vehicle
  • This step is the same as step 101 and will not be described in detail here.
  • Step 202 Perform grayscale processing on the image data to obtain a grayscale image of the image data of the object to be identified.
  • the original image is subjected to smoothing and grayscale processing.
  • the image is composed of pixels of different grayscale values, and the grayscale distribution in the image is one of the images.
  • An important feature, and the gray level histogram describes the gray level distribution in the image, which can intuitively show how much each gray level occupies in the image.
  • Step 203 Process the grayscale image by using a Sauber Sobel algorithm to obtain first differential image data of the image data.
  • the grayscale image obtained in step 202 is processed using the Sobel algorithm to obtain differential image data of the grayscale image, where the differential image data is a matrix corresponding to the grayscale image and Sobel. Matrix obtained by convolution of the matrix.
  • Step 204 Perform differential processing on the images in the image database corresponding to the object to be identified to obtain second differential image data.
  • the image database includes historical image data obtained by the object to be identified within a preset time period.
  • the above-mentioned three-level image is obtained by shooting similar scenes or recognition objects within a preset time period, and the original images obtained multiple times are stored in the image database for obtaining the first time Two differential images.
  • Step 205 By normalizing the second difference image data, a difference value distribution and a difference value cumulative distribution of the second difference image data are obtained.
  • the grayscale image obtained above is subjected to multiple differential processing experiments to obtain multiple second differential images. If 10 differential image experiments are performed, and the three-valued experiments with different thresholds are obtained, such as The four images with acceptable effects are taken out by the technician's naked eyes as shown in FIG. 5. Among them, it can be seen that the corresponding edges in the four images are clearly visible, so the four images are selected as the three values that meet the preset standards of the technicians. image.
  • the second difference image data corresponding to the above four images are selected for normalization processing to obtain the difference value distribution figure 6 (a) and the difference value cumulative distribution figure 6 of the second difference image ( b).
  • Step 206 Obtain the ratio of the separation noise and the signal in the second difference image data according to the difference value distribution and the difference value cumulative distribution, and determine it as a ternary coefficient.
  • the Sobel operator processes forward edge pixels
  • the curve in FIG. 6 (a) represents three values.
  • the ratio of the coefficient is between 0-1
  • the proportional relationship between each pixel value and the maximum pixel value max, and the cumulative distribution of each pixel value in the differential image shown in Figure 6 (b) is mapped in Figure 6 (a)
  • the display is shown as multiple vertical lines, showing the correspondence between the cumulative distribution of different pixels and the distribution of pixel differential values;
  • the original image is processed in the same scene, and the number of times the image is three-valued is determined by the actual situation, and the three-valued image that meets the preset standards is selected by the visual inspection of a technician, where the preset The standard is set by the technician according to the specific needs, for example, less noise, clearer edges, etc., and then obtain the histogram of the selected image, that is, the cumulative distribution probability of the difference value of the pixel data shown in Figure 6 (b) It can be known that the cumulative pixel distribution probability of the four images described in FIG. 5 is between 97% and 99%.
  • the ternary coefficient can be used to directly perform ternary image processing to obtain the most appropriate ternary image.
  • it further includes:
  • Step A1 performing grayscale processing on an image in an image database corresponding to the object to be identified to obtain a grayscale image of the image data;
  • This step is similar to step 202 and will not be described in detail here.
  • step A2 the grayscale image is subjected to three-level processing to obtain a plurality of test three-level images including pixel data of the object to be identified.
  • the three-valued image is composed of several lines in the horizontal direction, and each line is distributed with several pixel values on the three-valued image.
  • FIG. 8 it is a 145x385 three-valued image.
  • FIG. 9 there are 145 horizontal lines in the three-valued image, as shown in FIG. 9, assuming that the positive boundary pixel values distributed on one of the horizontal lines are 6-12, then for a road, each road boundary There are 3-6 pixels, which is the number of horizontal pixels.
  • Step A3 Determine the number of pixels in each of the test three-valued images according to a preset number of pixels and a preset pixel appearance probability
  • the existence probability of a road boundary is 50%
  • 98.4% is the preset distribution probability.
  • Step A4 Obtain a cumulative cumulative probability of difference values of the pixel data according to the number of pixels in the test ternary image.
  • a corresponding positive boundary pixel (a pixel with a positive value) or a negative boundary pixel (a pixel with a negative value) is extracted, and it is similar to FIG. 6 Histogram distribution.
  • the differential value distribution and the cumulative probability of the differential value of the pixel data of the differential image data are obtained.
  • Step A5 Obtain a ternary coefficient according to the cumulative distribution probability of the difference value.
  • the corresponding ratio that is, the ternary coefficient, can be found in the differential value distribution map of the pixel data.
  • step 207 a ternary image is obtained by applying the ternary threshold on the differential image data.
  • step 105 This step is the same as step 105 and will not be described in detail here.
  • the methods provided above are not limited to road recognition, but can also be applied to object recognition.
  • the original image is first taken through smoothing, grayscale, and differential processing. Then obtain a differential image, and then obtain a corresponding histogram based on the differential image, that is, the differential value distribution and the cumulative probability of the differential value of the pixel data of the differential image data. According to experience, the cumulative distribution probability is between 96% -99%.
  • the corresponding three-valued coefficient (ratio) is between 0.1-0.4, then the above-mentioned gray-scale image is subjected to three-valued processing according to the three-valued coefficient, and four three-valued images are obtained.
  • the technician can The relative three-value image is taken out for further recognition of traffic cones.
  • image data of the object to be identified is obtained, and the image data is subjected to graying processing to obtain a grayscale image of the image data of the object to be identified.
  • An algorithm processes the grayscale image, obtains first differential image data of the image data, performs differential processing on an image in an image database corresponding to the object to be identified, and obtains second differential image data;
  • the difference image data is normalized to obtain a difference value distribution and a difference value cumulative distribution of the second difference image data; and obtain a separation noise sum in the second difference image data according to the difference value distribution and the difference value cumulative distribution.
  • the ratio of the signals is determined as the ternary coefficient, which achieves the purpose of efficiently and systematically acquiring the ternary image of the original image.
  • FIG. 11 is a structural block diagram of a ternary image acquisition device according to an embodiment of the present invention.
  • the device includes:
  • An identification object image acquisition module 301 configured to acquire image data of an object to be identified during the driving of the vehicle;
  • the ternary coefficient obtaining module 303 is configured to obtain a ternary coefficient by performing preset processing on an image in an image database corresponding to the object to be identified;
  • a ternary threshold acquisition module 304 configured to obtain a ternary threshold according to a product of the ternary coefficient and a maximum pixel value in the first difference image data
  • a tri-level image acquisition module 305 is configured to obtain a tri-level image by applying the tri-level threshold on the differential image data.
  • the device includes:
  • An identification object image acquisition module 301 configured to acquire image data of an object to be identified during the driving of the vehicle;
  • the differential image data acquisition module 302 includes:
  • a grayscale image acquisition submodule 3021 configured to perform grayscale processing on the image data to obtain a grayscale image of the image data of the object to be identified;
  • a differential image data acquisition submodule 3022 is configured to process the grayscale image by using a Sauber Sobel algorithm to obtain first differential image data of the image data.
  • the ternary coefficient obtaining module 303 is configured to obtain a ternary coefficient by performing preset processing on an image in an image database corresponding to the object to be identified;
  • the ternary coefficient acquisition module includes:
  • a second differential image data acquisition submodule 3031 configured to perform differential processing on an image in an image database corresponding to the object to be identified to obtain second differential image data
  • a difference distribution acquisition submodule 3032 configured to obtain a difference value distribution and a difference value cumulative distribution of the second difference image data by normalizing the second difference image data
  • the trinization coefficient acquisition sub-module 3033 is configured to obtain a ratio of a separation noise and a signal in the second differential image data according to the difference value distribution and the cumulative difference value distribution, and determine the ratio as a trinization coefficient.
  • the ternary coefficient obtaining module 303 includes:
  • a grayscale image acquisition submodule configured to perform grayscale processing on an image in an image database corresponding to the object to be identified to obtain a grayscale image of the image data
  • a test sub-module configured to perform a three-value processing on the grayscale image to obtain a plurality of test three-valued images including pixel data of the object to be identified;
  • a pixel number acquisition sub-module configured to determine the number of pixels in each of the test three-valued images according to a preset number of pixels and a preset pixel appearance probability
  • a distribution value acquisition submodule configured to acquire a cumulative distribution probability of difference values of the pixel data according to the number of pixels in the test ternary image
  • the ternization coefficient acquisition submodule is configured to acquire a ternization coefficient according to the cumulative distribution probability of the difference value.
  • a ternary threshold acquisition module 304 configured to obtain a ternary threshold according to a product of the ternary coefficient and a maximum pixel value in the first difference image data
  • a tri-level image acquisition module 305 is configured to obtain a tri-level image by applying the tri-level threshold on the differential image data.
  • the image database includes historical image data obtained by the object to be identified within a preset time period.
  • an image acquisition module for the identification object is used to acquire image data of the object to be identified during the driving of the vehicle; a differential image data acquisition module is used to acquire the first differential image data of the image data; A ternary coefficient acquisition module is used to obtain a ternary coefficient by performing preset processing on an image in an image database corresponding to the object to be identified; a ternary threshold acquisition module is used to obtain the ternary coefficient A product of the maximum pixel value in the first difference image data to obtain a three-valued threshold value; a three-value image acquisition module is configured to obtain a three-valued value by applying the three-valued threshold value to the difference image data Into the image. It achieves the purpose of effectively and systematically obtaining the three-valued image of the original image.
  • An embodiment of the present invention further includes a vehicle, and the vehicle includes any three-valued image acquisition method or device in Embodiments 1 to 3.
  • the device embodiments described above are only schematic, and the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located One place, or it can be distributed across multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the objective of the solution of this embodiment. Those of ordinary skill in the art can understand and implement without creative labor.
  • the various component embodiments of the present invention may be implemented by hardware, or by software modules running on one or more processors, or by a combination thereof.
  • a microprocessor or a digital signal processor (DSP) may be used in practice to implement some or all functions of some or all components in a computing processing device according to an embodiment of the present invention.
  • the invention may also be implemented as a device or device program (e.g., a computer program and a computer program product) for performing part or all of the method described herein.
  • Such a program that implements the present invention may be stored on a computer-readable medium or may have the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.
  • Fig. 13 shows a computing processing device that can implement the method according to the invention.
  • the computing processing device traditionally includes a processor 1010 and a computer program product or computer-readable medium in the form of a memory 1020.
  • the memory 1020 may be an electronic memory such as a flash memory, an EEPROM (Electrically Erasable Programmable Read Only Memory), an EPROM, a hard disk, or a ROM.
  • the memory 1020 has a storage space 1030 of program code 1031 for performing any of the method steps in the above method.
  • the storage space 1030 for program code may include respective program codes 1031 respectively for implementing various steps in the above method. These program codes can be read from or written into one or more computer program products.
  • Such computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks.
  • Such a computer program product is typically a portable or fixed storage unit as described with reference to FIG. 14.
  • the storage unit may have a storage segment, a storage space, and the like arranged similar to the memory 1020 in the computing processing device of FIG. 13.
  • the program code may be compressed, for example, in a suitable form.
  • the storage unit includes computer-readable code 1031 ', that is, code that can be read by, for example, a processor such as 1010, and these codes, when run by a computing processing device, cause the computing processing device to execute the method described above Steps.
  • one embodiment means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Also, please note that the word examples "in one embodiment” herein do not necessarily refer to the same embodiment.
  • any reference signs placed between parentheses shall not be construed as limiting the claim.
  • the word “comprising” does not exclude the presence of elements or steps not listed in a claim.
  • the word “a” or “an” preceding an element does not exclude the presence of a plurality of such elements.
  • the invention can be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claim listing several devices, several of these devices may be embodied by the same hardware item.
  • the use of the words first, second, and third does not imply any order. These words can be interpreted as names.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

本发明提供了一种基于视觉摄像机的自阴影物体边缘识别方法、装置及车辆，所述方法包括：在车辆行驶过程中，采集待识别对象的影像数据，并进行微分处理；根据预设阈值对所述微分处理后的影像进行三值化处理，以获取包括正负向边界像素的三值化影像；根据所述正负向边界像素，获取表示所述待识别物的边界走势的正负向直线段；如果所述直线段创建了具备预设条件目标对象，则确定所述直线段为所述待识别对象的外围边界，否则根据所述预设条件所述确定第一边界；根据所述第一边界确定第二边界；通过所述第一边界和所述第二边界的位置，绘制所述待识别对象的外围边界。解决了现有技术中没有系统有效的方法识别自阴影物体边缘的问题。

Description

一种三值图像获取方法、装置及车辆
本申请要求在2018年06月29日提交中国专利局、申请号为201810711921.7发明名称为“一种三值图像获取方法、装置及车辆”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本发明涉及自动驾驶技术领域，包括一种三值图像获取方法、装置及车辆。
背景技术
ADAS(Advanced Driver Assistant Systems，高级驾驶辅助系统)可以利用安装在车上的各式各样传感器，在汽车行驶过程中随时来感应周围的环境，收集数据，并辨识、追踪不同运动状态的物体。再结合导航仪地图数据，进行系统的运算与分析，从而预先让驾驶者察觉到可能发生的危险，有效增加汽车驾驶的舒适性和安全性，尤其是在自动驾驶领域，ADAS中的各种传感器，如视觉相机具有像人类一样获得二维图像的功能，所以可作为自动驾驶车辆中ADAS中的主要传感器。
其中，视觉相机使用图像处理功能来捕捉车辆驾驶中的道路图像，并辅助驾驶员在自动驾驶模式下准确的识别道路，是自动驾驶领域的一项重要课题。而最关键的视觉相机图像处理技术为边缘提取，而通过获取检测对象的三值图更能有效检测图像中的边缘信息，通过边缘提取捕获图像中的物体和环境信息。
目前，尽管边缘提取是视觉相机的关键技术，但是并没有有效和系统的方法来获取效果较好的原始图像的三值图像，从而进行边缘图像的提取。
发明内容
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它目的、特征和优点能够更明显易懂，以下特举本发明的具体实施方式。
为解决上述问题，本发明的技术方案是这样实现的：
一种三值图像获取方法，所述方法包括：在车辆获取所述影像数据的第一差分图像数据；
通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；
根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；
通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
进一步的，所述获取所述影像数据的第一差分图像数据的步骤，包括：
对所述影像数据进行灰度化处理，获取所述待识别对象的影像数据的灰度图；
通过索伯Sobel算法处理所述灰度图，获取所述影像数据的第一差分图像数据。
进一步的，所述通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数的步骤，包括：
对对应所述待识别对象的图像数据库中的图像进行差分处理，获得第二差分图像数据；
通过对所述第二差分图像数据进行归一化，获取所述第二差分图像数据的差分值分布和差分值累计分布；
根据所述差分值分布和所述差分值累计分布获取第二差分图像数据中的分离噪声和信号的比率，确定为三值化系数。
进一步的，所述通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数的步骤，包括：
对对应所述待识别对象的图像数据库中的图像进行灰度处理，得到所述图像数据的灰度图；
将所述灰度图进行三值化处理，得到包括所述待识别对象的像素数据的多个测试三值图像；
根据所述测试三值图像中的像素数获取所述像素数据的差分值累计分布概率；
根据所述差分值累计分布概率获取三值化系数。
进一步的，所述图像数据库包括所述待识别对象在预设时间段内获取的历史图像数据。
相对于现有技术，本发明所述的三值图像获取方法具有以下优势：通过在车辆行驶过程中，获取待识别对象的影像数据；获取所述影像数据的第一差分图像数据；通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；通过所述三值化阈值在所述差分图像 数据上的应用，得到三值化图像。具有系统有效的获取原始图像的三值化图像的的有益效果。
本发明的另一目的在于提出一种三值图像获取装置，所述装置包括：识别对象影像获取模块，用于在车辆行驶过程中，获取待识别对象的影像数据；
差分图像数据获取模块，用于获取所述影像数据的第一差分图像数据；
三值化阈值获取模块，用于根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；
三值图像获取模块，用于通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
进一步的，所述差分图像数据获取模块，包括：
灰度图获取子模块，用于对所述影像数据进行灰度化处理，获取所述待识别对象的影像数据的灰度图；
差分图像数据获取子模块，用于通过索伯Sobel算法处理所述灰度图，获取所述影像数据的第一差分图像数据。
进一步的，所述三值化系数获取模块，包括：
第二差分图像数据获取子模块，用于对对应所述待识别对象的图像数据库中的图像进行差分处理，获得第二差分图像数据；
差分分布获取子模块，用于通过对所述第二差分图像数据进行归一化，获取所述第二差分图像数据的差分值分布和差分值累计分布；
三值化系数获取子模块，用于根据所述差分值分布和所述差分值累计分布获取第二差分图像数据中的分离噪声和信号的比率，确定为三值化系数。
进一步的，所述三值化系数获取模块，包括：
灰度图获取子模块，用于对对应所述待识别对象的图像数据库中的图像进行灰度处理，得到所述图像数据的灰度图；
测试子模块，用于将所述灰度图进行三值化处理，得到包括所述待识别对象的像素数据的多个测试三值图像；
像素数获取子模块，用于根据预设像素数与预设像素出现概率确定各所述测试三值图像中的像素数；
分布值获取子模块，用于根据所述测试三值图像中的像素数获取所述像素数据的差分值累计分布概率；
三值化系数获取子模块，用于根据所述差分值累计分布概率获取三值化系数。
进一步的，所述图像数据库包括所述待识别对象在预设时间段内获取的历史图像数据。
所述一种三值图像获取装置与上述一种三值图像获取方法相对于现有技术所具有的优势相同，在此不再赘述。
本发明的另一目的在于提出一种车辆，包括上述描述的三值图像获取方法及装置。
本发明的再一目的在于提出一种计算机程序，包括计算机可读代码，当所述计算机可读代码在计算处理设备上运行时，导致所述计算处理设备执行根据三值图像获取方法。
本发明的又一目的在于提出一种计算机可读介质，其中存储了上述计算机程序。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本发明实施例一所述的一种三值图像获取方法的流程图；
图2为本发明实施例中的图像处理结果示意图；
图3为本发明实施例中的图像处理结果示意图；
图4为本发明实施例中的图像处理原理示意图；
图5为本发明实施例中的图像处理结果示意图；
图6为本发明实施例中的图像处理原理示意图；
图7为本发明实施例二所述的一种三值图像获取方法的流程图；
图8为本发明实施例中的图像处理原理示意图；
图9为本发明实施例中的图像处理原理示意图；
图10为本发明实施例中的图像处理流程示意图；
图11为本发明实施例三所述的一种三值图像获取装置的结构框图；
图12为本发明实施例三所述的一种三值图像获取装置的结构框图；
图13为本发明实施例所述的一种计算机处理设备的结构示意图；
图14为本发明实施例所述的一种存储单元的结构示意图。
具体实施方式
下面将参照附图更详细地描述本公开的示例性实施例。虽然附图中显示 了本公开的示例性实施例，然而应当理解，可以以各种形式实现本公开而不应被这里阐述的实施例所限制。相反，提供这些实施例是为了能够更透彻地理解本公开，并且能够将本公开的范围完整的传达给本领域的技术人员。
需要说明的是，在不冲突的情况下，本发明中的实施例及实施例中的特征可以相互组合。
下面将参考附图并结合实施例来详细说明本发明。
实施例一
参照图1，为本发明实施例所述的一种三值图像获取方法的流程图，具体可以包括如下步骤：
步骤101，在车辆行驶过程中，获取待识别对象的影像数据。
在本发明实施例中，利用车辆装置的视觉系统采集车辆行驶中的待识别对象。
例如，获取的待识别对象如图2所示。图2可以是从由视觉系统中装置摄像头获取到的全局影像，其中白色方框中标示的图像范围为当前车辆主要识别对象为行驶道路，为提升后期影像处理的效率，如图3所示，后期图像处理主要针对该白色方框中的区域图像。
步骤102，获取所述影像数据的第一差分图像数据。
在本发明实施例中，可以通过Sobel滤波器(Filter)对该影像(Image)进行影像处理，获取第一差分图像数据。如图3所示，通过对待处理图像经过再采样、平滑处理后，通过Sobel滤波器对该图像进行处理得到差分图像数据(differential image)。
具体地，利用公式(1)中显示的Sobel算子，
Figure PCTCN2019093770-appb-000001
如图4所示，Raster Scan(光栅扫描)在水平方向上扫过影像上的每一个像素点，并对原影像上的XXX(m,n)变换为微分影像差分图像数据ddd(m,n)，具体可以参考如下代码：
Figure PCTCN2019093770-appb-000002
Figure PCTCN2019093770-appb-000003
步骤103，通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数。
在本发明实施例中，具体地，利用公式(2)，对该差分图像进行三值化处理：
Figure PCTCN2019093770-appb-000004
本发明实施例中，具体地，首先通过多次对待识别对象的图像数据进行不同阈值的三值化处理，获得最优识别结果经验，并根据该经验获得三值图像数据的正向边界像素或负向边界像素的累计分布概率的范围。再根据公式(2)获取上述差分图像数据的正向边界像素的差分值分布图，以及正向边界像素或负向边界像素的累计分布概率，并反向推断出对应上述经验的累计分布概率的范围的三值化系数。
例如，如图5所示的是多个差分图中的负向边界像素得到的多个三值化图像，根据研究人员的测试和观察，这四幅图分别符合预设条件，即体现出了噪点较少且边界清晰的特性，且上述四幅图对应的正向边界像素的累计分布概率范围为97％-99％之间，该概率分布范围即为三值化系数。
所以，如图6所示的，获取图3所示的差分图对应的正向边界像素的差分值分布图(图6上半部分)，以及正向边界像素负向边界像素的累计分布概率(图6下半部分)，根据之前获取的预设累计分布概率(threshold)为97％-99％之间，最优的那么在图6上半部分所示的差分值分布图中，可以获取针对累计分布概率为97％-99％对应的三值化系数(ratio)为0.1-0.4。
步骤104，根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值。
在本发明实施例中，将获取的三值化系数与获取的第一差分图像数据中的最大像素值相乘，得到的即是三值化阈值。如图6所示的，三值化系数若为0.3，那么在图6(a)中，根据最大像素值系数的直方图分布，即可以得到对应的三值化阈值时99％。
步骤105，通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
在本发明实施例中，通过已获得的三值化系数对上述得到的差分图像数据进一步进行三值化处理，得到该差分图像数据的三值图像。
具体可以参考如下三值化影像ttt(m,n)代码,以获取三值化影像ttt(m,n)：
Figure PCTCN2019093770-appb-000005
经过上述三值化影像处理，获取的三值化影像ttt(m,n)中每一像素点的取值从集合[-1,0,1]中选取。由此，差分影像ddd(m,n)中大于预设阈值Threshold的像素点被赋值为1，其中Threshold的取值为ratio*max(|ddd(：)|)差分影像ttt(m,n)中小于阈值Threshold的像素点被赋值为-1，其余像素点被赋值为0。这样，就可以区分出值为1的正向边界像素，值为-1的负向边界像素，而除边界像素以外的其他像素均被赋值为0。
例如，根据步骤103获得的三值化系数(ratio)在0.1-0.4之间，那么根据该三值化系数分别获取ratio分别取值为0.1、0.2、0.3、0.4获得效果较好的三值化图像。
在本发明实施例中，通过在车辆行驶过程中，获取待识别对象的影像数据；获取所述影像数据的第一差分图像数据；通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。达到了有效和系统的获取效果较好的原始图像的三值图像的目的。
实施例二
参照图7，为本发明实施例所述的一种三值图像获取方法的流程图，具 体可以包括如下步骤：
步骤201，在车辆行驶过程中，获取待识别对象的影像数据；
此步骤与步骤101相同，在此不再详述。
步骤202，对所述影像数据进行灰度化处理，获取所述待识别对象的影像数据的灰度图。
在本发明实施例中，如图3所示的，对原始图像经过平滑、灰度化处理，具体地，图像由不同灰度值的像素组成，图像中灰度的分布情况是该图像的一个重要特征，而灰度直方图就描述了图像中灰度分布情况，能够很直观的展示出图像中各个灰度级所占的多少。
步骤203，通过索伯Sobel算法处理所述灰度图，获取所述影像数据的第一差分图像数据。
在本发明实施例中，根据步骤102中描述的，利用Sobel算法处理步骤202中得到的灰度图，获取该灰度图像的差分图像数据，其中差分图像数据是灰度图对应的矩阵与Sobel矩阵进行卷积得到的矩阵。
步骤204，对对应所述待识别对象的图像数据库中的图像进行差分处理，获得第二差分图像数据。
其中，所述图像数据库包括所述待识别对象在预设时间段内获取的历史图像数据。
在本发明实施例中，上述三值化图像是经过对类似场景或识别对象在预设时间段内进行拍摄获取的，而多次获取的原始图像时存储在图像数据库中，以供随时获取第二差分图像。
步骤205，通过对所述第二差分图像数据进行归一化，获取所述第二差分图像数据的差分值分布和差分值累计分布。
在本发明实施例中，将上述得到的灰度图进行多次差分处理试验，得到多个第二差分图像，如果进行了10次差分图像试验，并通过不同阈值的三值化实验，得到如图5所示的通过技术人员的肉眼观察取出其中效果尚可的四幅图，其中，可以看出该四幅图中对应的边缘清晰可见，所以选择这四幅图为满足技术人员预设标准的三值图像。
所以，如图6所示的，挑选上述四幅图中对应的对第二差分图像数据进行归一化处理，获取第二差分图像的差分值分布图6(a)和差分值累计分布图6(b)。
步骤206，根据所述差分值分布和所述差分值累计分布获取第二差分图 像数据中的分离噪声和信号的比率，确定为三值化系数。
本发明实施例中，根据图6中，Sobel算子处理的是正向边缘像素，图像中的总像素NN＝55825，最大像素值max＝975，图6(a)中的曲线表示的是三值化系数ratio在0-1之间时，各像素值与最大像素值max之间的比例关系，且图6(b)所示的差分图像中各像素值的累计分布映射在图6(a)中显示为多个竖线，展示出了不同像素累计分布与像素差分值分布的对应关系；
在实际应用中，在同一场景下对待原始图像进行处理，并且进行图像三值化处理的次数取决于实际情况，并通过技术人员的肉眼检测挑选出符合预设标准的三值图像，其中预设标准由技术人员根据具体需求进行设置，例如，噪点较少，边缘较为清晰等，然后获取上述选择出的图像的直方图，即图6(b)中所示的像素数据的差分值累计分布概率，即可得知图5描述的四幅图像的像素累计分布概率在97％-99％之间。
根据确定的像素累计分布概率，确定对应的三值化系数为0.3左右较为合适。在之后的三值化图像处理中，如果是相同拍摄场景或者相同的识别对象，即可使用该三值化系数直接进行三值化图像处理，得到最为合适的三值化图像。
优选的，还包括：
步骤A1，对对应所述待识别对象的图像数据库中的图像进行灰度处理，得到所述图像数据的灰度图；
此步骤与步骤202相似，在此不再详述。
步骤A2，将所述灰度图进行三值化处理，得到包括所述待识别对象的像素数据的多个测试三值图像。
在本发明实施例中，假设三值图像由水平方向的若干线条组成，那么每一个线条上分布有该三值图像上的若干像素值，如图8所示的，是一个145x385的三值图像，那么该三值图像中存在的水平线为145条，如图9所示的，假设其中一条水平线上分布的正向边界像素值为6-12个，那么对于一条道路来说，每条道路边界上有3-6个像素点，即水平像素数。
步骤A3，根据预设像素数与预设像素出现概率确定各所述测试三值图像中的像素数；
在本发明实施例中，再假设道路边界的存在概率是50％，那么总像素数为12*0.5*145＝870，像素的累计分布概率为1-870/(145*385)＝98.4％， 也就是说98.4％是预设分布概率。
步骤A4，根据所述测试三值图像中的像素数获取所述像素数据的差分值累计分布概率；
在本发明实施例中，在上述得到的差分图像数据矩阵中，提取对应的正向边界像素(为正值的像素)或负向边界像素(为负值的像素)，获得其与图6相似的直方图分布。根据该差分图像数据的直方图分析，得到该差分图像数据的像素数据的差分值分布和差分值累计分布概率。
步骤A5，根据所述差分值累计分布概率获取三值化系数。
在本发明实施例中，由于之前确定了98.4％为预设分布概率，那么对应该预设分布概率，可以在像素数据的差分值分布图中找到对应的ratio，即三值化系数。
步骤207，通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
此步骤与步骤105相同，在此不再详述。
在实际应用中，上述提供的方法不仅限于道路识别，还可应用于物体识别，如图10所示，如果对于道路中的交通锥标进行识别，首先拍摄原始图像经过平滑、灰度、差分处理后得到差分图像，再根据差分图像得到对应的直方图，即差分图像数据的像素数据的差分值分布和差分值累计分布概率，其中根据经验得知累计分布概率在96％-99％之间，那么对应的三值化系数(ratio)在0.1-0.4之间，那么根据该三值化系数对上述灰度图进行三值化处理，得到四个三值化图像，技术人员可根据实际需求在其中取出相对于的三值图像进行进一步的交通锥标的识别处理。
在本发明实施例中，在车辆行驶过程中，获取待识别对象的影像数据，对所述影像数据进行灰度化处理，获取所述待识别对象的影像数据的灰度图，通过索伯Sobel算法处理所述灰度图，获取所述影像数据的第一差分图像数据，对对应所述待识别对象的图像数据库中的图像进行差分处理，获得第二差分图像数据；通过对所述第二差分图像数据进行归一化，获取所述第二差分图像数据的差分值分布和差分值累计分布；根据所述差分值分布和所述差分值累计分布获取第二差分图像数据中的分离噪声和信号的比率，确定为三值化系数，达到了有效和系统的获取效果较好的原始图像的三值图像的目的。
实施例三
参照图11，为本发明实施例所述的一种三值图像获取装置的结构框图，所述装置包括：
识别对象影像获取模块301，用于在车辆行驶过程中，获取待识别对象的影像数据；
差分图像数据获取模块302，用于获取所述影像数据的第一差分图像数据；
三值化系数获取模块303，用于通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；
三值化阈值获取模块304，用于根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；
三值图像获取模块305，用于通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
参照图12，为本发明实施例所述的一种三值图像获取装置的结构框图，所述装置包括：
识别对象影像获取模块301，用于在车辆行驶过程中，获取待识别对象的影像数据；
差分图像数据获取模块302，用于获取所述影像数据的第一差分图像数据；
优选的，所述差分图像数据获取模块302，包括：
灰度图获取子模块3021，用于对所述影像数据进行灰度化处理，获取所述待识别对象的影像数据的灰度图；
差分图像数据获取子模块3022，用于通过索伯Sobel算法处理所述灰度图，获取所述影像数据的第一差分图像数据。
三值化系数获取模块303，用于通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；
优选的，所述三值化系数获取模块，包括：
第二差分图像数据获取子模块3031，用于对对应所述待识别对象的图像数据库中的图像进行差分处理，获得第二差分图像数据；
差分分布获取子模块3032，用于通过对所述第二差分图像数据进行归一化，获取所述第二差分图像数据的差分值分布和差分值累计分布；
三值化系数获取子模块3033，用于根据所述差分值分布和所述差分值累计分布获取第二差分图像数据中的分离噪声和信号的比率，确定为三值化系 数。
优选的，所述三值化系数获取模块303，包括：
灰度图获取子模块，用于对对应所述待识别对象的图像数据库中的图像进行灰度处理，得到所述图像数据的灰度图；
测试子模块，用于将所述灰度图进行三值化处理，得到包括所述待识别对象的像素数据的多个测试三值图像；
像素数获取子模块，用于根据预设像素数与预设像素出现概率确定各所述测试三值图像中的像素数；
分布值获取子模块，用于根据所述测试三值图像中的像素数获取所述像素数据的差分值累计分布概率；
三值化系数获取子模块，用于根据所述差分值累计分布概率获取三值化系数。
三值化阈值获取模块304，用于根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；
三值图像获取模块305，用于通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
优选的，所述图像数据库包括所述待识别对象在预设时间段内获取的历史图像数据。
在本发明实施例中，通过识别对象影像获取模块，用于在车辆行驶过程中，获取待识别对象的影像数据；差分图像数据获取模块，用于获取所述影像数据的第一差分图像数据；三值化系数获取模块，用于通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；三值化阈值获取模块，用于根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；三值图像获取模块，用于通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。达到了有效和系统的获取效果较好的原始图像的三值图像的目的。
本发明实施例还包括一种车辆，所述车辆包括实施例一至实施例三中的任一三值图像获取方法或装置。
以上所描述的装置实施例仅仅是示意性的，其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现 本实施例方案的目的。本领域普通技术人员在不付出创造性的劳动的情况下，即可以理解并实施。
本发明的各个部件实施例可以以硬件实现，或者以在一个或者多个处理器上运行的软件模块实现，或者以它们的组合实现。本领域的技术人员应当理解，可以在实践中使用微处理器或者数字信号处理器(DSP)来实现根据本发明实施例的计算处理设备中的一些或者全部部件的一些或者全部功能。本发明还可以实现为用于执行这里所描述的方法的一部分或者全部的设备或者装置程序(例如，计算机程序和计算机程序产品)。这样的实现本发明的程序可以存储在计算机可读介质上，或者可以具有一个或者多个信号的形式。这样的信号可以从因特网网站上下载得到，或者在载体信号上提供，或者以任何其他形式提供。
例如，图13示出了可以实现根据本发明的方法的计算处理设备。该计算处理设备传统上包括处理器1010和以存储器1020形式的计算机程序产品或者计算机可读介质。存储器1020可以是诸如闪存、EEPROM(电可擦除可编程只读存储器)、EPROM、硬盘或者ROM之类的电子存储器。存储器1020具有用于执行上述方法中的任何方法步骤的程序代码1031的存储空间1030。例如，用于程序代码的存储空间1030可以包括分别用于实现上面的方法中的各种步骤的各个程序代码1031。这些程序代码可以从一个或者多个计算机程序产品中读出或者写入到这一个或者多个计算机程序产品中。这些计算机程序产品包括诸如硬盘，紧致盘(CD)、存储卡或者软盘之类的程序代码载体。这样的计算机程序产品通常为如参考图14所述的便携式或者固定存储单元。该存储单元可以具有与图13的计算处理设备中的存储器1020类似布置的存储段、存储空间等。程序代码可以例如以适当形式进行压缩。通常，存储单元包括计算机可读代码1031’，即可以由例如诸如1010之类的处理器读取的代码，这些代码当由计算处理设备运行时，导致该计算处理设备执行上面所描述的方法中的各个步骤。
本文中所称的“一个实施例”、“实施例”或者“一个或者多个实施例”意味着，结合实施例描述的特定特征、结构或者特性包括在本发明的至少一个实施例中。此外，请注意，这里“在一个实施例中”的词语例子不一定全指同一个实施例。
在此处所提供的说明书中，说明了大量具体细节。然而，能够理解， 本发明的实施例可以在没有这些具体细节的情况下被实践。在一些实例中，并未详细示出公知的方法、结构和技术，以便不模糊对本说明书的理解。
在权利要求中，不应将位于括号之间的任何参考符号构造成对权利要求的限制。单词“包含”不排除存在未列在权利要求中的元件或步骤。位于元件之前的单词“一”或“一个”不排除存在多个这样的元件。本发明可以借助于包括有若干不同元件的硬件以及借助于适当编程的计算机来实现。在列举了若干装置的单元权利要求中，这些装置中的若干个可以是通过同一个硬件项来具体体现。单词第一、第二、以及第三等的使用不表示任何顺序。可将这些单词解释为名称。
最后应说明的是：以上实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的精神和范围。

Claims (13)

  1. 一种三值图像获取方法，其特征在于，所述方法包括：
    在车辆行驶过程中，获取待识别对象的影像数据；
    获取所述影像数据的第一差分图像数据；
    通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；
    根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；
    通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
  2. 根据权利要求1所述的三值图像获取方法，其特征在于，所述获取所述影像数据的第一差分图像数据的步骤，包括：
    对所述影像数据进行灰度化处理，获取所述待识别对象的影像数据的灰度图；
    通过索伯Sobel算法处理所述灰度图，获取所述影像数据的第一差分图像数据。
  3. 根据权利要求1所述的三值图像获取方法，其特征在于，所述通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数的步骤，包括：
    对对应所述待识别对象的图像数据库中的图像进行差分处理，获得第二差分图像数据；
    通过对所述第二差分图像数据进行归一化，获取所述第二差分图像数据的差分值分布和差分值累计分布；
    根据所述差分值分布和所述差分值累计分布获取第二差分图像数据中的分离噪声和信号的比率，确定为三值化系数。
  4. 根据权利要求1所述的三值图像获取方法，其特征在于，所述通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数的步骤，包括：
    对对应所述待识别对象的图像数据库中的图像进行灰度处理，得到所述图像数据的灰度图；
    将所述灰度图进行三值化处理，得到包括所述待识别对象的像素数据的多个测试三值图像；
    根据预设像素数与预设像素出现概率确定各所述测试三值图像中的像 素数；
    根据所述测试三值图像中的像素数获取所述像素数据的差分值累计分布概率；
    根据所述差分值累计分布概率获取三值化系数。
  5. 根据权利要求1所述的三值图像获取方法，其特征在于，所述图像数据库包括所述待识别对象在预设时间段内获取的历史图像数据。
  6. 一种三值图像获取装置，其特征在于，所述装置包括：
    识别对象影像获取模块，用于在车辆行驶过程中，获取待识别对象的影像数据；
    差分图像数据获取模块，用于获取所述影像数据的第一差分图像数据；
    三值化系数获取模块，用于通过对对应所述待识别对象的图像数据库中的图像进行预设处理，得到三值化系数；
    三值化阈值获取模块，用于根据所述三值化系数与所述第一差分图像数据中的最大像素值的乘积，得到三值化阈值；
    三值图像获取模块，用于通过所述三值化阈值在所述差分图像数据上的应用，得到三值化图像。
  7. 根据权利要求6所述的三值图像获取装置，其特征在于，所述差分图像数据获取模块，包括：
    灰度图获取子模块，用于对所述影像数据进行灰度化处理，获取所述待识别对象的影像数据的灰度图；
    差分图像数据获取子模块，用于通过索伯Sobel算法处理所述灰度图，获取所述影像数据的第一差分图像数据。
  8. 根据权利要求6所述的三值图像获取装置，其特征在于，所述三值化系数获取模块，包括：
    第二差分图像数据获取子模块，用于对对应所述待识别对象的图像数据库中的图像进行差分处理，获得第二差分图像数据；
    差分分布获取子模块，用于通过对所述第二差分图像数据进行归一化，获取所述第二差分图像数据的差分值分布和差分值累计分布；
    三值化系数获取子模块，用于根据所述差分值分布和所述差分值累计分布获取第二差分图像数据中的分离噪声和信号的比率，确定为三值化系数。
  9. 根据权利要求6所述的三值图像获取装置，其特征在于，所述三值化系数获取模块，包括：
    灰度图获取子模块，用于对对应所述待识别对象的图像数据库中的图像进行灰度处理，得到所述图像数据的灰度图；
    测试子模块，用于将所述灰度图进行三值化处理，得到包括所述待识别对象的像素数据的多个测试三值图像；
    像素数获取子模块，用于根据预设像素数与预设像素出现概率确定各所述测试三值图像中的像素数；
    分布值获取子模块，用于根据所述测试三值图像中的像素数获取所述像素数据的差分值累计分布概率；
    三值化系数获取子模块，用于根据所述差分值累计分布概率获取三值化系数。
  10. 根据权利要求6所述的三值图像获取装置，其特征在于，所述图像数据库包括所述待识别对象在预设时间段内获取的历史图像数据。
  11. 一种车辆，其特征在于，所述车辆包括权利要求1-10中的任一三值图像获取方法及装置。
  12. 一种计算机程序，包括计算机可读代码，当所述计算机可读代码在计算处理设备上运行时，导致所述计算处理设备执行根据权利要求1-5中的任一个所述的三值图像获取方法。
  13. 一种计算机可读介质，其中存储了如权利要求12所述的计算机程序。
PCT/CN2019/093770 2018-06-29 2019-06-28 一种三值图像获取方法、装置及车辆 Ceased WO2020001630A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP19824956.7A EP3800614A4 (en) 2018-06-29 2019-06-28 METHOD AND APPARATUS FOR GENERATION OF A TERNARY IMAGE AND VEHICLE
US17/256,844 US11861844B2 (en) 2018-06-29 2019-06-28 Ternary image obtaining method and apparatus, and vehicle

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810711921.7A CN109993761B (zh) 2018-06-29 2018-06-29 一种三值图像获取方法、装置及车辆
CN201810711921.7 2018-06-29

Publications (1)

Publication Number Publication Date
WO2020001630A1 true WO2020001630A1 (zh) 2020-01-02

Family

ID=67128757

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/093770 Ceased WO2020001630A1 (zh) 2018-06-29 2019-06-28 一种三值图像获取方法、装置及车辆

Country Status (4)

Country Link
US (1) US11861844B2 (zh)
EP (1) EP3800614A4 (zh)
CN (1) CN109993761B (zh)
WO (1) WO2020001630A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113658252B (zh) * 2021-05-17 2024-08-23 毫末智行科技有限公司 用于估计摄像头仰角的方法、介质、装置及该摄像头
CN114022424A (zh) * 2021-10-25 2022-02-08 广西电网有限责任公司电力科学研究院 一种基于三值化的小车循迹避障识别方法
CN116735605B (zh) * 2023-06-16 2026-03-20 中国联合网络通信集团有限公司 毛刺检测方法、设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101277365A (zh) * 2007-03-27 2008-10-01 佳能株式会社 图像处理方法和图像处理设备
CN101888469A (zh) * 2009-05-13 2010-11-17 富士通株式会社 图像处理方法和图像处理设备
CN104573704A (zh) * 2013-10-09 2015-04-29 爱信精机株式会社 眼部检测装置、方法
US20150213611A1 (en) * 2014-01-29 2015-07-30 Canon Kabushiki Kaisha Image processing apparatus that identifies image area, and image processing method

Family Cites Families (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3490490B2 (ja) * 1994-01-28 2004-01-26 株式会社東芝 パターン画像処理装置及び画像処理方法
JP3318248B2 (ja) * 1997-12-25 2002-08-26 シャープ株式会社 画質改善装置及び記録媒体
US7149675B2 (en) * 2001-03-09 2006-12-12 Intel Corporation System and method for automatically mapping state elements for equivalence verification
JP2006309650A (ja) * 2005-05-02 2006-11-09 Calsonic Kansei Corp ナンバー認識装置及方法
US8320702B2 (en) * 2006-09-28 2012-11-27 Jadak Technologies, Inc. System and method for reducing specular reflection
JP5686533B2 (ja) * 2009-07-10 2015-03-18 株式会社東芝 磁気共鳴イメージング装置、画像処理装置および画像処理方法
JP2011034178A (ja) * 2009-07-30 2011-02-17 Sony Corp 画像処理装置および画像処理方法、並びにプログラム
JP5436138B2 (ja) * 2009-10-19 2014-03-05 キヤノン株式会社 画像処理装置、画像処理方法およびプログラム
JP2011176748A (ja) 2010-02-25 2011-09-08 Sony Corp 画像処理装置および方法、並びにプログラム
CN102890780B (zh) * 2011-07-19 2015-07-22 富士通株式会社 图像处理装置和方法
CN105447489B (zh) * 2015-11-13 2018-11-16 浙江传媒学院 一种图片ocr识别系统的字符与背景粘连噪声消除方法
CN107463667A (zh) * 2017-08-02 2017-12-12 大连理工大学 基于邻居像素点的共生局部三值模式的图像检索方法
CN108681737B (zh) * 2018-05-25 2022-06-07 长安大学 一种复杂光照下图像特征提取方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101277365A (zh) * 2007-03-27 2008-10-01 佳能株式会社 图像处理方法和图像处理设备
CN101888469A (zh) * 2009-05-13 2010-11-17 富士通株式会社 图像处理方法和图像处理设备
CN104573704A (zh) * 2013-10-09 2015-04-29 爱信精机株式会社 眼部检测装置、方法
US20150213611A1 (en) * 2014-01-29 2015-07-30 Canon Kabushiki Kaisha Image processing apparatus that identifies image area, and image processing method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HONGLIANG LI: "Unsupervized Video Segmentation With Low Depth of Field", IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 31 December 2007 (2007-12-31), pages 1742 - 1751, XP011195136 *
See also references of EP3800614A4

Also Published As

Publication number Publication date
CN109993761A (zh) 2019-07-09
EP3800614A1 (en) 2021-04-07
US11861844B2 (en) 2024-01-02
US20210256273A1 (en) 2021-08-19
CN109993761B (zh) 2021-04-09
EP3800614A4 (en) 2022-03-23

Similar Documents

Publication Publication Date Title
EP3036730B1 (en) Traffic light detection
CN110705405A (zh) 目标标注的方法及装置
CN104657735A (zh) 车道线检测方法、系统、车道偏离预警方法及系统
CN111080661A (zh) 基于图像的直线检测方法、装置以及电子设备
US20170154234A1 (en) Information processing device, information processing method, computer-readable recording medium, and inspection system
EP1892666A1 (en) A method, an apparatus and a computer-readable medium for processing an image dataset
CN113228035A (zh) 用于对象识别的系统和方法
JP2009048629A (ja) 検出方法
WO2020001631A1 (zh) 基于视觉摄像机的自阴影物体边缘识别方法、装置及车辆
CN115083008A (zh) 运动目标检测方法、装置、设备及存储介质
JP6903477B2 (ja) 軌道識別装置、およびプログラム
WO2019085929A1 (zh) 图像处理方法及其装置、安全驾驶方法
WO2020001630A1 (zh) 一种三值图像获取方法、装置及车辆
CN101369312B (zh) 检测图像中交叉口的方法和设备
CN104915642A (zh) 前方车辆测距方法及装置
Öztürk et al. Computer vision-based lane detection and detection of vehicle, traffic sign, pedestrian using YOLOv5
KR101236223B1 (ko) 차선 검출 방법
CN108090425B (zh) 一种车道线检测方法、装置及终端
WO2019149213A1 (zh) 基于影像识别路锥的方法、装置、存储介质以及车辆
CN115841632B (zh) 输电线路提取方法、装置以及双目测距方法
CN112308801A (zh) 基于大数据图像采集的道路交通追踪管理方法及系统
KR101620425B1 (ko) 주변 환경 정보를 이용한 차선 인식 시스템 및 그 방법
CN110490033B (zh) 用于车道检测的图像处理方法及装置
CN117094974A (zh) 基于图像的缺陷检测方法、装置、电子设备及存储介质
CN112785550B (zh) 图像质量值确定方法、装置、存储介质及电子装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19824956

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2019824956

Country of ref document: EP

Effective date: 20201229