CN103338325A - Chassis image acquisition method based on panoramic camera - Google Patents
Chassis image acquisition method based on panoramic camera Download PDFInfo
- Publication number
- CN103338325A CN103338325A CN2013102383598A CN201310238359A CN103338325A CN 103338325 A CN103338325 A CN 103338325A CN 2013102383598 A CN2013102383598 A CN 2013102383598A CN 201310238359 A CN201310238359 A CN 201310238359A CN 103338325 A CN103338325 A CN 103338325A
- Authority
- CN
- China
- Prior art keywords
- image
- chassis
- algorithm
- registration
- panoramic 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000003384 imaging method Methods 0.000 claims abstract description 34
- 230000008569 process Effects 0.000 claims abstract description 11
- 230000009466 transformation Effects 0.000 claims abstract description 8
- 230000000694 effects Effects 0.000 claims abstract description 7
- 238000013519 translation Methods 0.000 claims abstract description 6
- 230000000007 visual effect Effects 0.000 claims description 30
- 238000013461 design Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 239000004744 fabric Substances 0.000 claims description 8
- 241000251468 Actinopterygii Species 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 238000001514 detection method Methods 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 238000007500 overflow downdraw method Methods 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000001186 cumulative effect Effects 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 238000007689 inspection Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 230000001976 improved effect Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000012876 topography Methods 0.000 description 2
- 241000226585 Antennaria plantaginifolia Species 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 241000287181 Sturnus vulgaris Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012634 optical imaging Methods 0.000 description 1
- 230000002688 persistence Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 238000004064 recycling Methods 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 230000035807 sensation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Landscapes
- Image Processing (AREA)
- Studio Devices (AREA)
Abstract
The invention relates to a chassis image acquisition method based on a panoramic camera. The method adopts a two-dimensional imaging system with king-sized field of view which is fixedly mounted below the ground; when a vehicle passes through at a relatively low speed, the imaging system performs consecutive shooting; after the vehicle leaves, a one-dimensional translation image sequence is acquired through projection transformation; the observation effect of each part of the chassis from various viewing angles is recorded by the image sequence; during jointing process, the stripes of various images in different location perpendicular to the driving direction of the vehicle are selected for jointing to restore information from various viewing angles which are recorded, and different result images from different viewing angles are obtained. Compared with the conventional imaging recording schemes such as linear array CCD (Charge Coupled Device) scanning imaging and area array camera imaging, the chassis image acquisition method has obvious advantages in various aspects including cost control, joint chassis image quality, user operation manner, and the like.
Description
Technical field
The present invention relates to the safety detection method of vehicle chassis, relate in particular to a kind of chassis image-pickup method based on panoramic camera.
Background technology
Along with continuous progress and the expanding economy of society, people's living standard is also in continuous improve, and meanwhile, various social contradications are also in continuous generation.The universal use of motor vehicles, become the terrorist and place the important tool that explosive is made car bomb, also be the contraband of offender's first-selection, the delivery vehicle of smuggled goods simultaneously, owing to check comparatively difficulty at the bottom of the car, become the main positions of concealing article.These conducts have proposed challenge to current social peace and stable big atmosphere.In order to safeguard the stable of society, protect the safety of people's life and property, relevant departments such as public security, anti-terrorism hit by persistence always and pulverize terrified, evil force, are safeguarding the country and people's safety.For guaranteeing the safety of various important offices and department, be very important to the inspection of vehicles passing in and out.And can not accomplish thoroughly for the inspection of vehicle at present, especially the inspection at the bottom of the car is difficult to carry out.
Utilize automobile chassis to carry out unlawful activities and often appear at the bigger important place of flow motor, it is particularly necessary that automobile is carried out safety inspection fast and efficiently.But, because automobile chassis is very little apart from the height on ground, therefore check very inconvenient.The simplest also is that the method for generally using is to utilize the reflection of mirror, manually checks at the bottom of the car.The security staff reaches mirror at the bottom of the car, utilize the reflection of mirror to check the place that human eye can't directly be seen, this mode operating efficiency is very low, and the reliability of the inspection also qualification with the inspection personnel is relevant, especially when vehicle flowrate is bigger, do not satisfy checking the requirement of speed.
In order to overcome these problems, utilize digital imaging apparatus to carry out safe examination system (UVIS at the bottom of the equipment-Che of chassis image automation collection, under vehicle inspection system) begin to appear on the market, obtain general image at the bottom of the car by digital photographing apparatus and image processing techniques, intuitively observe fast at screen through the staff, realization is to the quick observation on vehicular traffic chassis, judge to have or not difference, and cooperate the enforcement of various actuators to inspection and the interception of suspect vehicle.
This system is the rapidity of computer, repeatability, and the height of human eye vision is intelligent to combine with abstract, constitutes the Vision Builder for Automated Inspection of a complexity jointly.This machine vision inspection system provides a kind of method that car load is checked fast, and information can be preserved in digitized mode, thereby constitutes huge automobile information database.Along with improving constantly of domestic scientific and technological level, the product consumption of this machine vision technique is constantly being increased.
At present, safe examination system can be divided into two kinds of forms according to the difference of mounting means at the bottom of the car: fixed and packaged type.Fixed is that the visual apparatus that will be used for checking is fixed in ground, and this mode is applicable to the place that need carry out safety inspection for a long time to automobile.Movable type is more flexible, can install immediately when needing, and uses when surprise check is carried out at the important road outpost of the tax office, and main feature is flexibly, fast.
The vehicle chassis safe examination system that adopts in the market, mostly be the mode that adopts scan table, when vehicle process scan table, inner camera is taken chassis topography continuously, and splices, the complete chassis image of final formation, camera is taken car base map picture by a speculum, and when vehicle was sailed through the camera site, camera was taken the measurement bottom diagram picture that crosses continuously, when car passed through scan table fully, camera can obtain the image at the bottom of the continuous whole car.
Characteristics such as high-resolution, high sensitivity, location of pixels information are strong because the linear array CCD image transducer has, compact conformation and self-scanning thereof, thereby, in many instances, the scanning system of imaging is made of line array CCD, optical imaging system, computer data acquiring and treatment system at the bottom of the car.But line array CCD can only photograph delegation's image at every turn, i.e. striped, and when subject moved along the direction vertical with cell array, the line-scan digital camera continuous high speed was taken many stripeds and is formed one width of cloth-two dimensional image.Some parameter of line array CCD can arrange, and the different object of taking moving speed, can regulate the shooting speed of camera by " pixel clock " that camera is set.Velocity variations occurs in the process that the movement velocity of object is being taken, in order to obtain the better image quality, just needing increases corresponding control appliance, such as hardware devices such as velocity transducer and encoders, otherwise " smear " or " overlapping " phenomenon can occur.In addition, use more line array CCD at present and mostly be digital, its price is often also very high, so utilize line array CCD as the system of vision facilities, system cost can be very high.And when the direction change appearred in object, the image of its output often was difficult to proofread and correct by software.At present the system of this quasi-representative has: the UnderVehicle Monitoring System of German SecuScan company, U.S. ETECHS, Gatekeeper company, safe examination system scan image product at the bottom of the domestic fixed car of S-G2048C that the refreshing rich safety science and technology in Shandong is arranged, its principle all is to adopt line array CCD as the imaging sensing unit, to obtain the image of high-resolution.
Also there is at present part to utilize two-dimensional imaging CCD camera to carry out the camera that detects at the bottom of the car, but it is because very near apart from ground distance at the bottom of the car, common camera can not be observed at the bottom of the complete car, need polyphaser to carry out system works, situation at the bottom of the usually direct video observation car, be difficult to be spliced into a complete car base map picture, the information that obtains is comparatively at random, brings difficulty to accurately judging.Also have report to utilize the video camera of several installations arranged side by side to take simultaneously, carry out the image co-registration splicing simultaneously in vehicle forward direction and vehicle-width direction, difficulty is bigger, directly has influence on the effect of the image of last generation.
Wherein checkout facility at the bottom of the two dimensional image car of Zhongdun Safety Technology Development Co., Beijing's production is mainly used in the vehicle bottom imaging inspection under the stationary vehicle condition, also is by continuous video output, to checking at the bottom of the car technically.
Check system mainly is to utilize the video record function of camera to constitute video check system at the bottom of the car at the bottom of the online car, when automobile by the time, trigger camera and take.Because the distance apart from ground at the bottom of the car is very little, needs the multi-section camera usually, a camera can only photograph a part at the bottom of the car, thus transversely arranged by a plurality of cameras, in order to photograph all images of car bottom width degree.The video image that photographs can show in the different viewports of individual monitor simultaneously, can show the image that different cameral obtains by video switch with a display, this test mode based on video needs the screen that stares at that the security staff do not stop, do not note slightly, just suspicious item might be omitted, and this check system based on video does not provide the complete image at the bottom of the car, the security staff can't the overview car at the bottom of.But this system configuration is simple, and applications little at vehicle flowrate and that travel speed is slower is more.The typical case produces video check system at the bottom of the brilliant car that Uniforce Scurity Systems Ltd arranged.
After also having at present research (referring to University Of Science and Technology Of Tianjin's master thesis " based on safety detection method research at the bottom of the car of area array cameras the array ") light source of being engaged in based on the two dimensional image splicing of area array CCD to illuminate the chassis, the camera array is taken the chassis topography through flat mirror reflects, and sends to concatenation module and splice.In splicing, the sequence of pictures of sending from many cameras can be received by system, carry out horizontally-spliced (the picture splicing between the camera array) and longitudinal spliced (single camera is in the splicing of different time picture shot) to these pictures; Namely to finish work such as two-dimentional many images match, registration, fusion.The realization of this process is very complicated, and the image of last registration is difficult to reach practical function, does not therefore adopt the actual product of this scheme.
Summary of the invention
For the bottom potential safety hazard that dynamically checks vehicles fast, first purpose of the present invention provides a kind of chassis image-pickup method based on panoramic camera, after panorama chassis scan table left later by vehicle, can obtain a series of panoramic pictures at each position at the bottom of the corresponding car, and then every Zhang Quanjing picture carried out projective transformation, figure as a result to conversion splices again, finally obtain complete chassis image, this method has remarkable advantages in various aspects such as the car base map picture element amount of cost control, splicing, user's modes of operation.Second purpose of the present invention provides a kind of chassis image capturing system based on panoramic camera.
In order to realize first above-mentioned purpose, the present invention has adopted following technical scheme:
Chassis image-pickup method based on panoramic camera, this method has adopted the two-dimentional flake imaging system of ultra-large vision field, two-dimentional flake imaging system is fixedly mounted on below the ground level, in the time of above vehicle crosses with lower speed, take continuously, after vehicle crosses, obtain the image sequence of one dimension translation by projective transformation, described sequence of pictures is noted the observing effect at various visual angles of each part on chassis, in splicing, choose in each pictures band perpendicular to the diverse location of vehicle direct of travel and splice and to restore the various visual angles information that records, obtain the figure as a result of different visual angles.
As preferably, described splicing comprises the steps:
1) image preliminary treatment
Some basic operations that image is handled comprise denoising, edge extracting and figure image intensifying, set up the matching template of image or image is carried out conversion;
2) image registration
Seek a coordinate transform, the coordinate points of overlapped part between the image is aimed at;
3) image co-registration
Carry out the splicing of adjacent image alignment area, the cumulative errors that cause in the elimination multiple image overall situation splicing and the distortion phenomenon in picture registration zone, output panorama stitching image.
As preferably, described step 2) algorithm of image registration adopts based on the algorithm in zone or based on the algorithm of feature, referring to utilize the relation of gray scale between two images to determine the parameter of changes in coordinates between image based on the algorithm in zone, is to utilize obvious characteristic in the image to come conversion between the computed image based on the algorithm of feature.As preferred again, described algorithm based on the zone comprises based on the pixel registration Algorithm in space with based on the algorithm of frequency domain; Algorithm based on feature comprises Harris Corner Detection Algorithm and SIFT yardstick invariant features transfer algorithm.
As most preferably, described step 2) algorithm that adopts of image registration comprises the steps:
1. image preliminary treatment: the image preliminary treatment is to eliminate or reduce gray-scale deviation between the image subject to registration, makes process of image registration to carry out smoothly;
2. choose template: from reference picture, choose a zone as template in mode manually or automatically;
3. template matches: in image subject to registration, search for, find the position of reference picture template in this image according to similarity measurement.
As preferably, described step 3) fusion method adopts the direct method of average, weighted mean method or multiresolution analysis method.
As preferred again, described step 3) fusion method adopts weighted mean method, and image co-registration can be expressed as:
Wherein f is the image after merging, f
1(x, y) and f
2(x y) is two width of cloth images to be spliced; w
1And w
2Represent the weights of two images to be spliced respective pixel in the overlapping region, satisfy:
w
1+w
2＝1 w
1,w
2＞0。
As preferably, the design parameter of the fish eye lens camera lens of described two-dimentional flake imaging system is: focal length is 1.16mm, aperture F=2,180 ° of the angles of visual field, system modulation transfer function mtf value at the 200lp/mm place greater than 0.4.
As preferably, it is 0 °～360 ° of the orientation angles of visual field that specific projection angle is chosen in described shooting, 0 °～-90 ° of the pitching angles of visual field.
In order to realize second above-mentioned purpose, the present invention has adopted following technical scheme:
Based on the chassis image capturing system of panoramic camera, this system comprises two-dimentional flake imaging system and picture treatment system, and described picture treatment system adopts above-mentioned any described method of technical scheme to obtain the figure as a result of different visual angles.
The present invention has adopted the two-dimentional flake imaging system of ultra-large vision field, be fixedly mounted on below the ground level, in the time of above vehicle crosses with lower speed, take continuously, after vehicle crosses, the arrangement image sequence, after at first utilizing the picture distortion correction technique image deformation that big view field imaging is intrinsic to revise, choose the middle rectangular part of image, carry out the registration between image sequence, then the associated region in the selected image sequence is merged, form entire vehicle base map picture.
Compare with imaging system at the bottom of the common car, this project has following characteristics:
(1) no matter with general camera or wide-angle camera, all have the visual field size issue, when taking chassis, because the camera distance chassis is very near, the information that photographs is just more limited, so a lot of research project all uses a plurality of cameras to finish.After general camera installed, when vehicle was not horizontal or vertical crossing, two-dimensional migration appearred in sequence of pictures, and the splicing difficulty increases, and effect reduces.This project adopts panorama camera, the hemisphere imaging, and the visual angle broadness can photograph the image of whole car bottom width degree direction simultaneously with a camera, and it is warm only need to carry out image at the vehicle direct of travel, can obtain complete car base map picture.
(2) Chang Yong method for registering images mainly contain based on the zone matching process (comprising correlation method, phase method) and based on the matching process (such as the SIFT feature) of feature.And the image sequence that this project collects by panoramic video exists distortion and the problem of visual angle change, be not that general camera review sequence has only projective transformation or orthogonal transform, so usual way just is difficult to prove effective.
This project product is compared with present existing linear array CCD scanning imaging, a burst of row camera imaging of face record scheme, in various aspects such as the car base map picture element amount of cost control, splicing, user's modes of operation, has remarkable advantages.Utilize this system dynamically fast to form vehicle bottom diagram picture, and carry out fast directly perceived the observation in conjunction with the staff at screen and judge, vehicle bottom information is covered all at one glance, can prevent that dangerous vehicle from entering, danger Du outdoors, prevent trouble before it happens.
Description of drawings
Fig. 1 is the use state diagram of system of the present invention.
Central strip was selected schematic diagram when Fig. 2 was the right-angle view chassis.
Central strip was selected schematic diagram when Fig. 3 observed chassis for side-looking forward.
Fig. 4 is the chassis that is spliced according to the condition of facing figure as a result.
Fig. 5 is the chassis that is spliced according to the direction of observation condition to antero-lateral horn degree figure as a result.
Fig. 6 is for according to observe the direction condition is spliced at the bottom of the car chassis figure as a result from tailstock angle.
Fig. 7 is the t1 panorama sketch of chassis constantly.
Fig. 8 is the t1 perspective view of chassis constantly.
Fig. 9 is the t2 panorama sketch of chassis constantly.
Figure 10 is the t2 perspective view of chassis constantly.
Figure 11 is the schematic diagram of image splicing.
Figure 12 is the basic flow sheet of image splicing.
Embodiment
The chassis image capturing system based on panoramic camera as shown in Figure 1, this system comprises two-dimentional flake imaging system and picture treatment system, two-dimentional flake imaging system is fixedly mounted on below the ground level, in the time of above vehicle crosses with lower speed, take continuously, after vehicle crosses, obtain the image sequence of one dimension translation by projective transformation, described sequence of pictures is noted the observing effect at various visual angles of each part on chassis, the picture treatment system is chosen in each pictures band perpendicular to the diverse location of vehicle direct of travel and is spliced and can restore the various visual angles information that records in splicing, obtain the figure as a result of different visual angles.
Be to choose central strip to splice as Fig. 2, be equivalent to the right-angle view chassis, Fig. 3 chooses upper sid strip to splice, and is equivalent to side-looking forward and observes chassis.Fig. 4, Fig. 5, Fig. 6 are the image of same chassis, and Fig. 4 is spliced according to the condition of facing, and Fig. 5 is spliced according to the direction of observation condition to the antero-lateral horn degree, and Fig. 6 is spliced according to direction condition at the bottom of tailstock angle observation car.
Below concrete parameter and the method that the present invention relates to is described in detail, but the selection of concrete parameter and method is not limited to following parameter and method.
(1) panoramic imagery Design for optical system
For realize at the bottom of the car in the Polaroid covering of the one camera of overall width direction, project has adopted fish eye lens, can implement the monitoring of non-blind areas to the vertical zones more than 180 ° of 360 ° of levels.Because the angle of visual field is very big, in fish-eye design, for fear of visual field, edge illumination according to cos
4(ω) descend, introduce a large amount of barrel distortion, can reduce picture side's angle of half field-of view greatly like this, thereby guarantee the slow decline of edge illumination, improved the uniformity of the illumination of practising physiognomy.
Because fish eye lens has been introduced a large amount of barrel distortion, the shared image planes of imaging that the characteristics of panorama picture of fisheye lens show as the core visual field are big, and visual field, the edge shared image planes of part are little.Therefore, in order to improve the resolution of edge imaging, need analyze fish-eye design feature.
Fish-eye imaging is based on " non-similar " imaging thought, being the image mapping relations, (y is half image height by the QUOTE y=f * tan ω y=f * tan ω of conventional camera lens, f is focal length, and ω is angle of half field-of view) become different types of image mapping relations, mainly comprise following several:
(A) equidistant projection imaging relations, y=f * ω QUOTE y=f * ω
Solid angle projection relation such as (B),
QUOTE
Stereo projection relation such as (C),
QUOTE
(D) rectangular projection relation etc., y=f * sin (ω)
The most frequently used be designed to equidistant projection, analyze the relation of edge resolution and fish eye lens design parameter below with equidistant projection.
The angle of visual field is 180 ° in the lens design, adopts the image mapping relations of equidistant projection, and y=f * (1+ δ) * ω is then arranged, and wherein δ is the F-theta distortion.
If it is ω 1QUOTE ω 1 and ω 2 that edge's image height is respectively the angle of visual field of y1 and y2 correspondence, then have:
y1＝f×(1+δ)×ω1 y2＝f×(1+δ)×ω2
Two formulas are subtracted each other to be had: Δ y=f * (1+ δ) * Δ ω
The angular resolution at edge is: S=Δ y/ Δ ω=f * (1+ δ).
When Δ y one timing, Δ ω is more big, and then edge angle resolution is more big.Therefore, according to following formula as can be known, fish-eye focal length is more little, and edge resolution is more big, and when the f-theta distortion was pincushion distortion, the angular resolution at edge was also more big.
Therefore, it is shorter to design fish-eye focal length as much as possible, simultaneously, introduces positive f-theta distortion at the edge, can effectively improve edge resolution.
According to top analysis, the present invention has designed a fish eye lens, and at the mega pixel CMOS photoelectric sensor of 1/3rd inches image planes, pixel dimension is 2.2 μ m, and half image height of flake imaging is h=1.8mm.Adopt anti-long distance structural design, guarantee that camera lens has long back focal length.The design parameter of camera lens is: focal length is 1.16mm, aperture F=2, and 180 ° of the angles of visual field, system modulation transfer function mtf value greater than 0.4, therefore has good imaging resolution capability at the 200lp/mm place, can obtain the image of high-resolution, low distortion.
(2) the projection expansion of image and direction adjustment
The panorama camera as long as adjust the projective transformation deflection angle, just can obtain the photographic plate that vehicle vertically passes through owing to be the hemisphere imaging.By choosing specific projection angle, guarantee that image sequence does strict rigid motion, have only very simple translation to change.This can reduce the registration difficulty, can guarantee that again the direction of the car base map picture of last output is consistent.
As shown in Figure 7, be the t1 panorama sketch of chassis constantly.By selected projection angle, we can obtain up facing along the chassis long side direction, the end of from a part of image of chassis from this width of cloth panorama sketch, as shown in Figure 8.Can clearly see that the y plotted is the long limit of chassis from Fig. 8, the x X direction is the minor face of chassis, and this image is a front view of looking up at the bottom of car, is the part of chassis.Along with the movement of car, can from each width of cloth panorama sketch, obtain this front view at different chassis position.As Fig. 9 and Figure 10, be t2 panorama sketch and the perspective view of chassis constantly.As seen, Fig. 8 and Figure 10 after the different projections constantly have only the upper and lower translation motion basically, can realize registration, splicing at an easy rate.To sum up, according to the projection image sequence of these different parts, just can splice the image of whole chassis.
(3) splicing of projected image
Image splicing (Image Mosaic), be the advanced row space registration of image sequence that has lap each other with a group, comprise the wide visual angle of each image sequence or the technology of 360 degree visual angle panoramic pictures through forming a width of cloth after image conversion, resampling and the image co-registration again.
Showing as Figure 11, is the schematic diagram of an image splicing.
The method of image splicing is a lot, and different algorithm steps has certain difference, but process roughly is identical.In general, the image splicing mainly comprises following a few step:
1) image preliminary treatment
Such as some basic operations (as denoising, edge extracting, figure image intensifying) that image is handled, set up the matching template of image or image is carried out certain conversion etc.
2) image registration
The quality of image splicing mainly depends on the registration accuracy of image.The key problem machine of image registration is sought a coordinate transform exactly, and the coordinate points of overlapped part between the image is aimed at.Registration Algorithm should guarantee the precision of registration, makes amount of calculation be unlikely to excessive again, and this is the committed step of present image splicing.
3) image co-registration
Through image registration, accurately obtain after the spatial alternation relation between the image, need be fused into a big image or panorama sketch to many original images.This process mainly comprises: carry out the splicing of adjacent image alignment area, the cumulative errors that cause in the elimination multiple image overall situation splicing and the distortion phenomenon in picture registration zone, output panorama stitching image.
Figure 12 has provided the basic flow sheet of image splicing.
Image registration and image co-registration are the core technologies of image splicing.
The algorithm that image registration is adopted mainly contains two classes: a class is based on the algorithm in zone, refers to utilize the relation of gray scale between two images to determine the parameter of changes in coordinates between image, comprising the pixel registration Algorithm based on the space, based on the algorithm of frequency domain etc.The another kind of algorithm that is based on feature is to utilize obvious characteristic in the image to come conversion between the computed image, rather than utilizes information whole in the image, comprising the Harris Corner Detection Algorithm, and SIFT yardstick invariant features transfer algorithm etc.
Project has adopted the template matching method based on the zone, is the class methods in the method for registering images.The main something in common of these class methods is at first will choose template from reference picture, and the recycling similarity measurement finds the position of this template in image subject to registration, thereby obtains the transformation relation between the image.The key step of such algorithm is:
1) image preliminary treatment: the image preliminary treatment is to eliminate or reduce gray-scale deviation between the image subject to registration, makes process of image registration to carry out smoothly.
2) choose template: from reference picture, choose a zone as template in mode manually or automatically.
3) template matches: in image subject to registration, search for, find the position of reference picture template in this image according to similarity measurement.
With the closely-related major issue of image registration be exactly image co-registration.Image co-registration be with informix in two width of cloth images in piece image and with the method for visualizing technique for displaying.The algorithm of image co-registration should satisfy: the border transition of image in splicing wanted nature, can splice seam by removal of images; Make loss original image information minimizing as far as possible in the image splicing, the blending algorithm of image will be applicable to the character of institute's stitching image.Main image interfusion method has: the direct method of average, weighted mean method, multiresolution analysis method.
Image-region after the present invention adopts the linear weighted function method of average to last coupling merges, but traditional linear weighted function fusion method will merge whole coincidence zones, and this had both increased workload, is easy to generate again and merges error (as ghost).Project approach is improved this problem, at different image shift, reasonably selects integration region, reduces corresponding circle of sensation in order to be effective basis as far as possible, has so both reduced amount of calculation, has improved effect again.
The weighted mean method image co-registration can be expressed as:
Wherein f is the image after merging, f
1(x, y) and f
2(x y) is two width of cloth images to be spliced.w
1And w
2Represent the weights of two images to be spliced respective pixel in the overlapping region, satisfy:
w
1+w
2＝1 w
1,w
2＞0
Triangle (cap) function method that R.Szeliski mentions in its classical paper is to weight w
1And w
2Arrange, make that in every width of cloth edge of image place weight be 0, be close to 1 near the center weight.
Claims (10)
1. based on the chassis image-pickup method of panoramic camera, it is characterized in that this method has adopted the two-dimentional flake imaging system of ultra-large vision field, two-dimentional flake imaging system is fixedly mounted on below the ground level, in the time of above vehicle crosses with lower speed, take continuously, after vehicle crosses, obtain the image sequence of one dimension translation by projective transformation, described sequence of pictures is noted the observing effect at various visual angles of each part on chassis, in splicing, choose in each pictures band perpendicular to the diverse location of vehicle direct of travel and splice and to restore the various visual angles information that records, obtain the figure as a result of different visual angles.
2. the chassis image-pickup method based on panoramic camera according to claim 1 is characterized in that described splicing comprises the steps:
1) image preliminary treatment
Some basic operations that image is handled comprise denoising, edge extracting and figure image intensifying, set up the matching template of image or image is carried out conversion;
2) image registration
Seek a coordinate transform, the coordinate points of overlapped part between the image is aimed at;
3) image co-registration
Carry out the splicing of adjacent image alignment area, the cumulative errors that cause in the elimination multiple image overall situation splicing and the distortion phenomenon in picture registration zone, output panorama stitching image.
3. the chassis image-pickup method based on panoramic camera according to claim 2, it is characterized in that step 2) algorithm of image registration adopts based on the algorithm in zone or based on the algorithm of feature, referring to utilize the relation of gray scale between two images to determine the parameter of changes in coordinates between image based on the algorithm in zone, is to utilize obvious characteristic in the image to come conversion between the computed image based on the algorithm of feature.
4. the chassis image-pickup method based on panoramic camera according to claim 3 is characterized in that algorithm based on the zone comprises based on the pixel registration Algorithm in space with based on the algorithm of frequency domain; Algorithm based on feature comprises Harris Corner Detection Algorithm and SIFT yardstick invariant features transfer algorithm.
5. the chassis image-pickup method based on panoramic camera according to claim 3 is characterized in that step 2) algorithm that adopts of image registration comprises the steps:
1. image preliminary treatment: the image preliminary treatment is to eliminate or reduce gray-scale deviation between the image subject to registration, makes process of image registration to carry out smoothly;
2. choose template: from reference picture, choose a zone as template in mode manually or automatically;
3. template matches: in image subject to registration, search for, find the position of reference picture template in this image according to similarity measurement.
6. the chassis image-pickup method based on panoramic camera according to claim 2 is characterized in that the step 3) fusion method adopts the direct method of average, weighted mean method or multiresolution analysis method.
7. the chassis image-pickup method based on panoramic camera according to claim 7 is characterized in that the step 3) fusion method adopts weighted mean method, and image co-registration can be expressed as:
Wherein f is the image after merging, f
1(x, y) and f
2(x y) is two width of cloth images to be spliced; w
1And w
2Represent the weights of two images to be spliced respective pixel in the overlapping region, satisfy:
w
1+w
2＝1 w
1,w
2＞0。
8. the chassis image-pickup method based on panoramic camera according to claim 1, the design parameter that it is characterized in that the fish eye lens camera lens of two-dimentional flake imaging system is: focal length is 1.16mm, aperture F=2,180 ° of the angles of visual field, system modulation transfer function mtf value at the 200lp/mm place greater than 0.4.
9. the chassis image-pickup method based on panoramic camera according to claim 1 is characterized in that it is 0 °～360 ° of the orientation angles of visual field that specific projection angle is chosen in shooting, 0 °～-90 ° of the pitching angles of visual field.
10. based on the chassis image capturing system of panoramic camera, it is characterized in that this system comprises two-dimentional flake imaging system and picture treatment system, described picture treatment system adopts the described method of claim 1～9 to obtain the figure as a result of different visual angles.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102383598A CN103338325A (en) | 2013-06-14 | 2013-06-14 | Chassis image acquisition method based on panoramic camera |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102383598A CN103338325A (en) | 2013-06-14 | 2013-06-14 | Chassis image acquisition method based on panoramic camera |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103338325A true CN103338325A (en) | 2013-10-02 |
Family
ID=49246415
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013102383598A Pending CN103338325A (en) | 2013-06-14 | 2013-06-14 | Chassis image acquisition method based on panoramic camera |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103338325A (en) |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103986912A (en) * | 2014-05-21 | 2014-08-13 | 南京大学 | Double-direction real-time vehicle chassis image synthetic method based on civil IPC |
| CN104463886A (en) * | 2014-12-18 | 2015-03-25 | 西南交通大学 | Processing method and device for images shot by line-scan digital camera |
| CN104463235A (en) * | 2014-11-18 | 2015-03-25 | 中国铁道科学研究院电子计算技术研究所 | Fault recognition method and device based on operation images of motor train unit |
| CN104967769A (en) * | 2015-07-16 | 2015-10-07 | 深圳中安高科电子有限公司 | Vehicle bottom scanning system and method |
| CN105430240A (en) * | 2015-06-30 | 2016-03-23 | 广州市科灵电子工程有限公司 | Vehicle chassis photographing device |
| WO2016107478A1 (en) * | 2014-12-30 | 2016-07-07 | 清华大学 | Vehicle chassis inspection method and system |
| WO2016107474A1 (en) * | 2014-12-30 | 2016-07-07 | 清华大学 | Vehicle checking method and system |
| CN105814886A (en) * | 2013-12-16 | 2016-07-27 | 索尼公司 | Image processing device, image processing method and program |
| CN107402221A (en) * | 2017-08-08 | 2017-11-28 | 广东工业大学 | A kind of defects of display panel recognition methods and system based on machine vision |
| CN107808392A (en) * | 2017-10-31 | 2018-03-16 | 中科信达（福建）科技发展有限公司 | The automatic method for tracking and positioning of safety check vehicle and system of open scene |
| CN107871346A (en) * | 2016-12-19 | 2018-04-03 | 珠海市杰理科技股份有限公司 | driving recorder |
| CN107872616A (en) * | 2016-12-19 | 2018-04-03 | 珠海市杰理科技股份有限公司 | Vehicle recording method and device |
| CN107944504A (en) * | 2017-12-14 | 2018-04-20 | 北京木业邦科技有限公司 | Plank identifies and machine learning method, device and the electronic equipment of plank identification |
| CN108352057A (en) * | 2015-11-12 | 2018-07-31 | 罗伯特·博世有限公司 | Vehicle camera system with multi-camera alignment |
| CN108475437A (en) * | 2015-04-10 | 2018-08-31 | 邦迪克斯商用车系统有限责任公司 | 360 ° of viewing systems of vehicle of video camera, calibration system and method are placed with corner |
| CN109348114A (en) * | 2018-11-26 | 2019-02-15 | Oppo广东移动通信有限公司 | Imaging device and electronic apparatus |
| CN109981985A (en) * | 2019-03-29 | 2019-07-05 | 上海智觅智能科技有限公司 | A kind of continuous stitching algorithm of double vision frequency |
| CN111076822A (en) * | 2019-12-13 | 2020-04-28 | 富德康(北京)科技股份有限公司 | Splicing array method for generating bulk thermodynamic diagram |
| CN111242847A (en) * | 2020-01-10 | 2020-06-05 | 上海西井信息科技有限公司 | Gateway-based image splicing method, system, equipment and storage medium |
| CN112488995A (en) * | 2020-11-18 | 2021-03-12 | 成都主导软件技术有限公司 | Intelligent injury judging method and system for automatic train maintenance |
| CN114240759A (en) * | 2021-12-29 | 2022-03-25 | 交通运输部公路科学研究所 | Vehicle bottom imaging processing method and device |
| CN116743942A (en) * | 2023-07-17 | 2023-09-12 | 盛视科技股份有限公司 | A two-way splicing vehicle bottom imaging method based on camera array |
| CN117221681A (en) * | 2023-09-20 | 2023-12-12 | 浙江舜宇智领技术有限公司 | Vehicle bottom environment detection camera, vehicle bottom environment detection method and vehicle system |
| CN118521671A (en) * | 2024-07-19 | 2024-08-20 | 深圳中安高科电子有限公司 | CMOS area array camera array train bottom imaging method and device |
| WO2025060771A1 (en) * | 2023-09-20 | 2025-03-27 | 浙江舜宇智领技术有限公司 | Vehicle bottom environment test camera, vehicle bottom environment test method, and vehicle system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE20311529U1 (en) * | 2003-07-26 | 2003-10-23 | Linhart, Jiri, 94469 Deggendorf | Video mirror for detecting hidden objects e.g. under vehicles has video cameras instead of mirrors and consists of rod, monitor and housing, which has integrated video cameras, microphone, sensors and metal detector |
| CN101009819A (en) * | 2006-12-27 | 2007-08-01 | 湖南大学 | Vehicle security vision intelligent detection device |
| CN101945257A (en) * | 2010-08-27 | 2011-01-12 | 南京大学 | Synthesis method for extracting chassis image of vehicle based on monitoring video content |
-
2013
- 2013-06-14 CN CN2013102383598A patent/CN103338325A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE20311529U1 (en) * | 2003-07-26 | 2003-10-23 | Linhart, Jiri, 94469 Deggendorf | Video mirror for detecting hidden objects e.g. under vehicles has video cameras instead of mirrors and consists of rod, monitor and housing, which has integrated video cameras, microphone, sensors and metal detector |
| CN101009819A (en) * | 2006-12-27 | 2007-08-01 | 湖南大学 | Vehicle security vision intelligent detection device |
| CN101945257A (en) * | 2010-08-27 | 2011-01-12 | 南京大学 | Synthesis method for extracting chassis image of vehicle based on monitoring video content |
Cited By (39)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105814886A (en) * | 2013-12-16 | 2016-07-27 | 索尼公司 | Image processing device, image processing method and program |
| CN105814886B (en) * | 2013-12-16 | 2020-09-15 | 索尼公司 | Image processing apparatus, image processing method and program |
| CN103986912A (en) * | 2014-05-21 | 2014-08-13 | 南京大学 | Double-direction real-time vehicle chassis image synthetic method based on civil IPC |
| CN103986912B (en) * | 2014-05-21 | 2017-04-12 | 南京大学 | Two-way real-time vehicle chassis image synthesis method based on civilian IPC |
| CN104463235A (en) * | 2014-11-18 | 2015-03-25 | 中国铁道科学研究院电子计算技术研究所 | Fault recognition method and device based on operation images of motor train unit |
| CN104463235B (en) * | 2014-11-18 | 2019-01-18 | 中国铁道科学研究院电子计算技术研究所 | Fault recognition method and device based on EMU operation image |
| CN104463886A (en) * | 2014-12-18 | 2015-03-25 | 西南交通大学 | Processing method and device for images shot by line-scan digital camera |
| WO2016107474A1 (en) * | 2014-12-30 | 2016-07-07 | 清华大学 | Vehicle checking method and system |
| CN105809655B (en) * | 2014-12-30 | 2021-06-29 | 清华大学 | Vehicle inspection method and system |
| CN105809655A (en) * | 2014-12-30 | 2016-07-27 | 清华大学 | Vehicle checking method and system |
| US10289699B2 (en) | 2014-12-30 | 2019-05-14 | Tsinghua University | Vehicle inspection methods and systems |
| WO2016107478A1 (en) * | 2014-12-30 | 2016-07-07 | 清华大学 | Vehicle chassis inspection method and system |
| CN108475437B (en) * | 2015-04-10 | 2021-12-03 | 邦迪克斯商用车系统有限责任公司 | 360 degree look-around system for vehicle with corner-mounted camera, calibration system and method |
| CN108475437A (en) * | 2015-04-10 | 2018-08-31 | 邦迪克斯商用车系统有限责任公司 | 360 ° of viewing systems of vehicle of video camera, calibration system and method are placed with corner |
| CN105430240A (en) * | 2015-06-30 | 2016-03-23 | 广州市科灵电子工程有限公司 | Vehicle chassis photographing device |
| CN104967769A (en) * | 2015-07-16 | 2015-10-07 | 深圳中安高科电子有限公司 | Vehicle bottom scanning system and method |
| CN108352057B (en) * | 2015-11-12 | 2021-12-31 | 罗伯特·博世有限公司 | Vehicle camera system with multi-camera alignment |
| CN108352057A (en) * | 2015-11-12 | 2018-07-31 | 罗伯特·博世有限公司 | Vehicle camera system with multi-camera alignment |
| CN107871346A (en) * | 2016-12-19 | 2018-04-03 | 珠海市杰理科技股份有限公司 | driving recorder |
| CN107872616A (en) * | 2016-12-19 | 2018-04-03 | 珠海市杰理科技股份有限公司 | Vehicle recording method and device |
| CN107871346B (en) * | 2016-12-19 | 2020-10-27 | 珠海市杰理科技股份有限公司 | driving recorder |
| CN107872616B (en) * | 2016-12-19 | 2021-03-19 | 珠海市杰理科技股份有限公司 | Driving recording method and device |
| CN107402221A (en) * | 2017-08-08 | 2017-11-28 | 广东工业大学 | A kind of defects of display panel recognition methods and system based on machine vision |
| CN107808392B (en) * | 2017-10-31 | 2023-04-07 | 中科信达（福建）科技发展有限公司 | Automatic tracking and positioning method and system for security check vehicle in open scene |
| CN107808392A (en) * | 2017-10-31 | 2018-03-16 | 中科信达（福建）科技发展有限公司 | The automatic method for tracking and positioning of safety check vehicle and system of open scene |
| CN107944504A (en) * | 2017-12-14 | 2018-04-20 | 北京木业邦科技有限公司 | Plank identifies and machine learning method, device and the electronic equipment of plank identification |
| CN107944504B (en) * | 2017-12-14 | 2024-04-16 | 北京木业邦科技有限公司 | Board recognition and machine learning method and device for board recognition and electronic equipment |
| CN109348114A (en) * | 2018-11-26 | 2019-02-15 | Oppo广东移动通信有限公司 | Imaging device and electronic apparatus |
| CN109981985A (en) * | 2019-03-29 | 2019-07-05 | 上海智觅智能科技有限公司 | A kind of continuous stitching algorithm of double vision frequency |
| CN111076822A (en) * | 2019-12-13 | 2020-04-28 | 富德康(北京)科技股份有限公司 | Splicing array method for generating bulk thermodynamic diagram |
| CN111242847B (en) * | 2020-01-10 | 2021-03-30 | 上海西井信息科技有限公司 | Gateway-based image stitching method, system, device and storage medium |
| CN111242847A (en) * | 2020-01-10 | 2020-06-05 | 上海西井信息科技有限公司 | Gateway-based image splicing method, system, equipment and storage medium |
| CN112488995A (en) * | 2020-11-18 | 2021-03-12 | 成都主导软件技术有限公司 | Intelligent injury judging method and system for automatic train maintenance |
| CN112488995B (en) * | 2020-11-18 | 2023-12-12 | 成都主导软件技术有限公司 | Intelligent damage assessment method and system for automated train maintenance |
| CN114240759A (en) * | 2021-12-29 | 2022-03-25 | 交通运输部公路科学研究所 | Vehicle bottom imaging processing method and device |
| CN116743942A (en) * | 2023-07-17 | 2023-09-12 | 盛视科技股份有限公司 | A two-way splicing vehicle bottom imaging method based on camera array |
| CN117221681A (en) * | 2023-09-20 | 2023-12-12 | 浙江舜宇智领技术有限公司 | Vehicle bottom environment detection camera, vehicle bottom environment detection method and vehicle system |
| WO2025060771A1 (en) * | 2023-09-20 | 2025-03-27 | 浙江舜宇智领技术有限公司 | Vehicle bottom environment test camera, vehicle bottom environment test method, and vehicle system |
| CN118521671A (en) * | 2024-07-19 | 2024-08-20 | 深圳中安高科电子有限公司 | CMOS area array camera array train bottom imaging method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103338325A (en) | Chassis image acquisition method based on panoramic camera | |
| TWI554976B (en) | Surveillance systems and image processing methods thereof | |
| US6677982B1 (en) | Method for three dimensional spatial panorama formation | |
| CN101814181B (en) | Unfolding method for restoration of fisheye image | |
| US20100148066A1 (en) | Thermal imaging camera for taking thermographic images | |
| US8848035B2 (en) | Device for generating three dimensional surface models of moving objects | |
| JPWO2011096251A1 (en) | Stereo camera | |
| KR101759798B1 (en) | Method, device and system for generating an indoor two dimensional plan view image | |
| US11689811B2 (en) | Method and apparatus for obtaining enhanced resolution images | |
| US11978222B2 (en) | Three-dimensional light field technology-based optical unmanned aerial vehicle monitoring system | |
| CN109360150A (en) | A kind of joining method and device of the panorama depth map based on depth camera | |
| Hong et al. | Full parallax three-dimensional display from Kinect v1 and v2 | |
| KR102498028B1 (en) | Surveillance Camera Systems and Mothod of Using the Same | |
| Cao et al. | Generating panoramic unfolded image from borehole video acquired through APBT | |
| JP5628068B2 (en) | Tunnel wall development image acquisition system | |
| CN109636763A (en) | A kind of intelligence compound eye monitoring system | |
| CN107135336A (en) | A kind of video camera array | |
| KR101452342B1 (en) | Surveillance Camera Unit And Method of Operating The Same | |
| Song et al. | Robust 3D reconstruction with omni-directional camera based on structure from motion | |
| JP2010256296A (en) | Omnidirectional three-dimensional space recognition input apparatus | |
| CN108810426A (en) | Infrared large-view-field splicing method and system | |
| CN108492254B (en) | Image acquisition system and method | |
| Zhu et al. | 3D Measurements in cargo inspection with a gamma-ray linear pushbroom stereo system | |
| JP3253328B2 (en) | Distance video input processing method | |
| CN106131534A (en) | Omnibearing stereo camera head and system and method thereof |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20131002 |
|
| RJ01 | Rejection of invention patent application after publication |