CN103049738A - Method for segmenting multiple vehicles connected through shadows in video - Google Patents
Method for segmenting multiple vehicles connected through shadows in video Download PDFInfo
- Publication number
- CN103049738A CN103049738A CN2012105258129A CN201210525812A CN103049738A CN 103049738 A CN103049738 A CN 103049738A CN 2012105258129 A CN2012105258129 A CN 2012105258129A CN 201210525812 A CN201210525812 A CN 201210525812A CN 103049738 A CN103049738 A CN 103049738A
- Authority
- CN
- China
- Prior art keywords
- image
- blob
- piece
- mentioned
- sub
- 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.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a method for segmenting multiple vehicles connected through shadows in a video. The method comprises the following steps of: performing edge computing on a vehicle area of an image; then, performing comparison and judgment on images obtained after computing; removing outer contours of the shadows through edge information; and then, removing the shadows of the vehicles through a morphological method. As the edge information is mainly concentrated at the part of the outer contours of the shadows, the goal that the shadows are removed is achieved through image difference, and meanwhile, as the edge information of the vehicles is relatively rich, and moreover, the edge information is insensitive to illumination variations, the method has universality.
Description
Technical field
The present invention relates to computer vision and pattern-recognition, digital image processing field, particularly, relate to many Method of Vehicle Segmentations that shade connects in a kind of video.
Background technology
Because vehicle and motion shade thereof have very similar movement properties feature, so in process background difference, can be connected in one when extracting foreground target, if (shade as referred to herein all is the motion shade not eliminate shade, as follows) can have a huge impact subsequent operations such as vehicle detection and vehicle trackings, so the elimination of shade is necessary.Wherein, eliminate shade in the blob piece, situation about will run into is divided into two classes: the first kind is to only have a car and shade thereof in the blob piece, and Equations of The Second Kind is to have two (or more than) cars to connect together by shade in the blob piece.
At present, the algorithm of most existing all is to go to carry out the elimination of shade from two aspects.The first kind is based on the method for model, more commonly the HMM(mixed Gauss model), these class methods will be known illumination in advance, the priori conditions such as direction of traffic, relatively be suitable for specific scene, algorithm does not have universality, can not be applicable to simultaneously arbitrarily scene.Equations of The Second Kind is based on the method for attribute, and these class methods are utilized the attributive character of vehicle and shade itself, and universality is relatively good, and without any need for priori.
Method based on attribute has a lot, common are color, brightness, colourity, normalization color space, edge, texture, gradient.As select to attempt using the shades of colour space instead, such as HSV, YCbCr, HIS etc., and the attributes such as comprehensive utilization brightness, colourity, texture, gradient eliminate shade, but that the problem of running into is the dash area eliminated is incomplete, misses one's aim.As shown in Figure 1 and Figure 2, utilize brightness, texture and gradient attribute to eliminate the final effect figure of vehicle shadow, can find out the image after the processing, have the part shade above the vehicle.
Summary of the invention
The object of the invention is to, for the problems referred to above, propose many Method of Vehicle Segmentations that shade connects in a kind of video, to realize that can eliminate shade fully has advantages of universality simultaneously.
For achieving the above object, the technical solution used in the present invention is:
Many Method of Vehicle Segmentations that shade connects in a kind of video may further comprise the steps:
Step 1: input video frame, to the original image a in the input video frame, at first carry out the background calculus of differences, obtain the foreground image b of binaryzation (0 and 255);
Step 2: select vehicle region in above-mentioned image b, this zone divides the prospect that obtains to obtain by the consecutive image background subtraction; And the gray-scale value of part all set to 0 beyond this is regional, obtained image c; Confine carrying out the blob piece in the above-mentioned hot spot region, be a minimum boundary rectangle of blob piece, obtain image d;
Step 3: above-mentioned image d is carried out morphologic Kai ﹑ closed operation, obtain image e; According to the upper left side point coordinate value that the wide and above-mentioned blob piece of the long ﹑ of blob piece is confined, intercept among the original image a of above-mentioned former figure corresponding to this blob confine the part original image obtain image f;
Step 4: ask boundary operation to obtain edge image g with edge detection algorithm to above-mentioned image f, ask boundary operation to obtain the outward flange image h of two-value foreground picture corresponding to blob piece to above-mentioned image e, described image g and image h respective pixel are subtracted each other and are obtained image i;
Step 5: the blob piece among the above-mentioned image c is carried out vertical operation and levels operation, and image will be divided into x connected region; A described x connected region is designated as respectively m and n in the length of side number of and vertical direction horizontal along the blob piece, and wherein x is greater than 1;
Step 6: the size of judging m and n, if m＞n, the sub-blob piece of then selecting above-mentioned vertical operation to confine, and the sub-blob piece that this sub-blob piece and above-mentioned levels operation obtain compared, find above-mentioned two parts that sub-blob piece overlaps, and the sub-blob piece of levels operation is exported in the part that will overlap;
Step 7: if m＜n, then select the sub-blob piece of confining of above-mentioned levels operation, and the sub-blob piece that this sub-blob piece and above-mentioned vertical operation obtain compared, find above-mentioned two parts that sub-blob piece overlaps, and the sub-blob piece of vertical operation is exported in the part that will overlap.
According to a preferred embodiment of the invention, the morphologic Kai ﹑ closed operation in the step 3 is: the execution closing operation of mathematical morphology on the mathematical morphology and execution morphology opening operation;
Described execution closing operation of mathematical morphology: use same structural element first image to be carried out corroding operation after the expansive working;
Described execution morphology opening operation carries out expansive working after namely using same structural element first image to be corroded operation.
According to a preferred embodiment of the invention, described vertical operation and levels operation are specific as follows:
Described vertical operation refers to each pixel column corresponding to image, finds that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0;
Described levels operation refers to each pixel column corresponding to image, finds that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0.
According to a preferred embodiment of the invention, described background calculus of differences adopts edge detection algorithm.
According to a preferred embodiment of the invention, described edge detection algorithm comprises SOBEL operator and Canny operator.
Technical scheme of the present invention is eliminated the outline of shade by marginal information, then eliminate vehicle shadow by morphologic method, mainly concentrate on the outline part of shade because of marginal information, simultaneously because vehicle edge information is abundanter, and marginal information is insensitive to illumination variation, thereby realize eliminating the purpose of shade by the background difference, and because vehicle edge information is abundanter, so have universality.
Below by drawings and Examples, technical scheme of the present invention is described in further detail.
Description of drawings
Fig. 1 is the existing image that adopts before attribute carries out the shade elimination;
Fig. 2 is that Fig. 1 adopts the image after attribute carries out the shade elimination;
Fig. 3 is the Method of Vehicle Segmentation process flow diagram that shade connects in the described video of the embodiment of the invention;
Fig. 4 a to Fig. 4 n is the image that produces in the Method of Vehicle Segmentation process that shade connects in the described video of the embodiment of the invention;
Fig. 5 a to Fig. 5 c is the schematic diagram of putting the possibility situation when in the video two cars being arranged;
Fig. 6 a to Fig. 6 h is the schematic diagram of putting the possibility situation when in the video three cars being arranged;
Fig. 7 a to Fig. 7 o is the schematic diagram of putting the possibility situation when in the video four cars being arranged.
Embodiment
Below in conjunction with accompanying drawing the preferred embodiments of the present invention are described, should be appreciated that preferred embodiment described herein only is used for description and interpretation the present invention, is not intended to limit the present invention.
As shown in Figure 3, many Method of Vehicle Segmentations that shade connects in a kind of video may further comprise the steps:
Step 1: input video frame, to the original image a in the input video frame, at first carry out the background calculus of differences, obtain the foreground image b of binaryzation;
Step 2: select vehicle region in image b, this zone divides the prospect that obtains to obtain by the consecutive image background subtraction; And the gray-scale value of part all set to 0 beyond this is regional, obtained image c; Confine carrying out the blob piece in the above-mentioned hot spot region, be a minimum boundary rectangle of blob piece, obtain image d;
Step 3: above-mentioned image d is carried out morphologic Kai ﹑ closed operation, obtain image e; According to the upper left side point coordinate value that the wide and above-mentioned blob piece of the long ﹑ of blob piece is confined, intercept among the original image a of above-mentioned former figure corresponding to this blob confine the part original image obtain image f;
Step 4: ask boundary operation to obtain edge image g with edge detection algorithm to above-mentioned image f, ask boundary operation to obtain the outward flange image h of two-value foreground picture corresponding to blob piece to above-mentioned image e, image g and image h respective pixel are subtracted each other and are obtained image i;
Step 5: the blob piece among the above-mentioned image c is carried out vertical operation and levels operation, and image will be divided into x connected region; X connected region is designated as respectively m and n in the length of side number of and vertical direction horizontal along the blob piece, and wherein x is greater than 1;
Step 6: the size of judging m and n, if m＞n, the sub-blob piece of then selecting above-mentioned vertical operation to confine, and the sub-blob piece that this sub-blob piece and above-mentioned levels operation obtain compared, find above-mentioned two parts that sub-blob piece overlaps, and the sub-blob piece of levels operation is exported in the part that will overlap;
Step 7: if m＜n, then select the sub-blob piece of confining of above-mentioned levels operation, and the sub-blob piece that this sub-blob piece and above-mentioned vertical operation obtain compared, find above-mentioned two parts that sub-blob piece overlaps, and the sub-blob piece of vertical operation is exported in the part that will overlap.
Wherein, the morphologic Kai ﹑ closed operation in the step 3 is: the execution closing operation of mathematical morphology on the mathematical morphology and execution morphology opening operation; Carry out closing operation of mathematical morphology: use same structural element first image to be carried out corroding operation after the expansive working; Carry out the morphology opening operation, carry out expansive working after namely using same structural element first image to be corroded operation.The background calculus of differences adopts edge detection algorithm.Edge detection algorithm comprises SOBEL operator and Canny operator.
Vertical operation and levels operation are specific as follows:
Vertical operation refers to each pixel column corresponding to image, finds that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0;
Levels operation refers to each pixel column corresponding to image, finds that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0.
After these pretreatment operation, each blob piece is carried out rectangle confine.Actually method also do not require and prejudge out independent in each a blob piece car, Hai Shi many cars that are connected together by shade, what at first we introduced is how to eliminate shade when only having single vehicle.
Step a, image pre-service
1. background difference
For original image a, at first carry out the background difference, obtain the foreground image b of binaryzation.The method of background difference is not limit, and what use in an embodiment is the SOBS algorithm.
2. hot spot region and blob piece are demarcated
In image b, select the hot spot region, and the gray-scale value of part beyond this zone is all set to 0, obtain image c; Confine (a minimum boundary rectangle that is the blob piece) to carrying out the blob piece in the hot spot region, obtain image d.
3. morphological operation
Owing among the d a lot of holes being arranged, and often having some noises, so image d is carried out morphologic Kai ﹑ closed operation, obtaining image e; Wide according to the Chang ﹑ of blob piece De, and the upper left side point coordinate value of the minimum boundary rectangle of blob piece, intercept the original image (colour) corresponding to the minimum boundary rectangle part of this blob among the former figure a, obtain image f.
Step b, ask for marginal information
SOBEL Operator Method with classics is asked the edge to image f, then spends the outward flange of two-value foreground picture corresponding to blob piece that difference image e represents, obtains image i.
Step c, vertical operation and levels operation
Vertical operation and levels operation are similar, and vertical operation refers to each pixel column corresponding to image, find that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0; Levels operation refers to each pixel column corresponding to image, finds that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0.Through this two steps operation, image will be divided into a lot of connected regions.
Steps d, connected region are demarcated
The method that adopts connected domain to demarcate is demarcated respectively the connected region that forms through the image behind vertical operation and the levels operation, employing be the method that eight connected regions are demarcated.By demarcating the part that to distinguish each connection in the image, thereby calculate the area (referring to the pixel number here) of each connected domain, simultaneously can know clearly that each connected component accounts for the ratio of whole image, (this ratio can be regulated according to actual conditions to delete those smaller connected regions, generally can get 10%), obtain result images l and image m.
Step e, car body reconstruct
With (l) that obtain and (m) and (i) that obtain previously by pixel do AND operation (be the grey scale pixel value of correspondence position among image i, l and the m be at 255 o'clock then the grey scale pixel value of image n correspondence position also be made as 255, otherwise other pixel gray-scale value all sets to 0 among the image n) obtain image graph as n, at this moment just can carry out car body reconstruct.
The method of reconstruct is vertical operation to be obtained image o and levels operation obtains image p, then to through the result of vertical operation and levels operation by pixel do inclusive-OR operation (be the grey scale pixel value of correspondence position among image o and the image p have one be 255 o'clock then the grey scale pixel value of image s correspondence position also be made as 255, otherwise other pixel gray-scale value is all put (0) among the image s, just obtains the car body image s after the reconstruct.
It should be noted that if also contain the connected region of small size among image o and the image p, also will again demarcate with connected domain, then remove the less connected region of area by certain area ratio, obtain figure q and image r.
Step f, Output rusults
Because the object of operation is a blob piece, may be corresponding among the former figure a lot of blob pieces, so the blob piece that generates after the current reconstruct to be outputed on the bianry image by original coordinate position, as the input of subsequent operation.
At last, can realize that based on the blob piece shade of many cars is eliminated, also will use vertical operation and levels operation.Concrete method is: after carrying out the necessary pre-service such as denoising point, the blob piece is obtained connected domain number m by vertical operation first, obtain connected domain number n by levels operation again.If 1. m=n=1 then illustrates it is single vehicle, eliminate shade with the method for foregoing single vehicle; If 2. m〉n, then explanation is many cars, can adopt first vertical operation to carry out confining of a blob piece again, obtains m sub-blob, then in this m sub-blob with the method elimination shade of foregoing single vehicle; If 3. m＜n then illustrates it is many cars, can adopt first levels operation to carry out again confining of a blob piece, obtain n sub-blob, then the method with foregoing single vehicle is eliminated shade in this n sub-blob; If 4. m=n ≠ 1 then illustrates it is many cars, can use 2. or the operation of method 3..
Now all possible situation that links to each other by shade among the blob is sorted out.The front is mentioned, and is divided into generally three classes and considers.1. as the first kind, 2. with 4. as Equations of The Second Kind, 3. as the 3rd class (hypothesis is 4. all with 2. method operation, so can unify to become a class here).
Then will know the value that m and n are concrete, this is the necessary condition of judging.The front is mentioned, and there is no need to look for special algorithm to judge the number of vehicles that links to each other by shade among each blob, but directly comprise the situation that might occur in decision condition.Next will provide 2 cars with illustrated method, the situation that m and n might occur in 3 cars and 4 the car situations, and then provide mathematical formulae with the method for formal proof.
See among upper Fig. 5 a to Fig. 7 o:
1. during two cars, the value of m and n has 3 kinds of situations, and (m n) comprised that ((2 1) ﹑ (2 2) do not comprise (1 1) to 1 2) ﹑;
2. during three cars, the value of m and n has 8 kinds of situations, (m n) comprised that (((((((3 2) ﹑ (3 3) do not comprise (1 1) to 3 1) ﹑ to 2 3) ﹑ to 2 2) ﹑ to 2 1) ﹑ to 1 3) ﹑ to 1 2) ﹑;
3. during four cars, the value of m and n has 15 kinds of situations, (m n) comprised that (((((((((((((4 3) ﹑ (4 4) do not comprise (1 1) to 4 2) ﹑ to 4 1) ﹑ to 3 4) ﹑ to 3 3) ﹑ to 3 2) ﹑ to 3 1) ﹑ to 2 3) ﹑ (2 4) to 2 2) ﹑ to 2 1) ﹑ to 1 4) ﹑ to 1 3) ﹑ to 1 2) ﹑;
Formal proof:
Suppose variable k(k
) the expression number of vehicles, when k=2, have 3 kinds of situations, namely
Kind; When k=3, have 8 kinds of situations, namely
Kind; When k=4, have 15 kinds of situations, namely
Kind.
Derivation can obtain: when k=5, have 24 kinds of situations; When k=6, have 35 kinds of situations;
Therefore, can obtain the formalization formula:
K(k is arranged in blob
) car links to each other by shade, the value one of m and n is total
Plant possibility.
Through above analysis, can set Rule of judgment by the value of m ﹑ n.Need to prove: all represent with rectangle for the vehicle that makes things convenient for of expressing in the diagram, in fact, when the Xing of vehicle Zhuan ﹑ Da Xiao ﹑ travel direction had various variation, top graphic analysis also was that effectively these variations can't affect the judgement to m and n value.
Carried out more again confining of a second son blob piece according to m and n in the blob piece, in every sub-blob piece, operated, as being single vehicle in the fruit blob piece, the single vehicle disposal route of just speaking of with (); As still containing many cars in the fruit blob piece, extract with regard to again carrying out blob, until in the blob piece of operation a car is only arranged.The sure success of this method, but more consuming time.In the system of reality, consider that eliminating the topmost purpose of shade is to cut off shade, can only carry out a second son blob piece and extract.
In the shade of video is eliminated, at first image is carried out pre-service, then judge the blob number of blocks of confining after processing, if quantity is 0, then not processing finishes; If quantity is 1, then process according to the method for bicycle shade; If quantity is greater than 1, the method for many cars shade is processed.
In sum, the present invention also has following advantage:
(1) our algorithm is that the front and back frame is irrelevant, only relevant with present frame, is applicable to frame losing video flowing or frame-skipping video flowing;
(2) contact not tight with the background difference algorithm;
(3) do not need to regulate parameter according to different scenes, unique parameter is the classical threshold value of edge detection operator;
(4) algorithm is based on blob's but not based on (pixel-based) of pixel, only use some shirtsleeve operations, and can use multi-core CPU to accelerate, and can process in real time after the acceleration.
Morphology in the literary composition is mathematical morphology, its basic thought is to go to measure and extract correspondingly-shaped in the image to reach the purpose to graphical analysis and identification with the structural element with certain form, mathematical morphology is molecular by one group of morphologic algebraic operation, its fundamental operation has 4: expand (or expansion), corrosion (or erosion), opening and closing, they respectively have characteristics in bianry image and gray level image.
Carry out closing operation of mathematical morphology, namely use same structural element first image to be carried out corroding operation after the expansive working.
Carry out the morphology opening operation, carry out expansive working after namely using same structural element first image to be corroded operation.
It should be noted that at last: the above only is the preferred embodiments of the present invention, be not limited to the present invention, although with reference to previous embodiment the present invention is had been described in detail, for a person skilled in the art, it still can be made amendment to the technical scheme that aforementioned each embodiment puts down in writing, and perhaps part technical characterictic wherein is equal to replacement.Within the spirit and principles in the present invention all, any modification of doing, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (5)
1. many Method of Vehicle Segmentations that shade connects in the video is characterized in that, may further comprise the steps:
Step 1: input video frame, to the original image a in the input video frame, at first carry out the background calculus of differences, obtain the foreground image b of binaryzation;
Step 2: select vehicle region in above-mentioned image b, this zone divides the prospect that obtains to obtain by the consecutive image background subtraction; And the gray-scale value of part all set to 0 beyond this is regional, obtained image c; Confine carrying out the blob piece in the above-mentioned hot spot region, be a minimum boundary rectangle of blob piece, obtain image d;
Step 3: above-mentioned image d is carried out morphologic Kai ﹑ closed operation, obtain image e; According to the upper left side point coordinate value that the wide and above-mentioned blob piece of the long ﹑ of blob piece is confined, intercept among the original image a of above-mentioned former figure corresponding to this blob confine the part original image obtain image f;
Step 4: ask boundary operation to obtain edge image g with edge detection algorithm to above-mentioned image f, ask boundary operation to obtain the outward flange image h of two-value foreground picture corresponding to blob piece to above-mentioned image e, described image g and image h respective pixel are subtracted each other and are obtained image i;
Step 5: the blob piece among the above-mentioned image c is carried out vertical operation and levels operation, and image will be divided into x connected region; A described x connected region is designated as respectively m and n in the length of side number of and vertical direction horizontal along the blob piece, and wherein x is greater than 1;
Step 6: the size of judging m and n, if m＞n, the sub-blob piece of then selecting above-mentioned vertical operation to confine, and the sub-blob piece that this sub-blob piece and above-mentioned levels operation obtain compared, find above-mentioned two parts that sub-blob piece overlaps, and the sub-blob piece of levels operation is exported in the part that will overlap;
Step 7: if m＜n, then select the sub-blob piece of confining of above-mentioned levels operation, and the sub-blob piece that this sub-blob piece and above-mentioned vertical operation obtain compared, find above-mentioned two parts that sub-blob piece overlaps, and the sub-blob piece of vertical operation is exported in the part that will overlap.
2. many Method of Vehicle Segmentations that shade connects in the video according to claim 1 is characterized in that, the morphologic Kai ﹑ closed operation in the described step 3 is: the execution closing operation of mathematical morphology on the mathematical morphology and execution morphology opening operation;
Described execution closing operation of mathematical morphology: use same structural element first image to be carried out corroding operation after the expansive working;
Described execution morphology opening operation carries out expansive working after namely using same structural element first image to be corroded operation.
3. many Method of Vehicle Segmentations that shade connects in the video according to claim 2 is characterized in that, described vertical operation and levels operation are specific as follows:
Described vertical operation refers to each pixel column corresponding to image, finds that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0;
Described levels operation refers to each pixel column corresponding to image, finds that first is 255 point with last gray-scale value in these row, and the gray scale of then putting the point between them is 255 entirely, and other gray scales are set to 0.
4. many Method of Vehicle Segmentations that shade connects according to claim 2 or in the 3 described videos is characterized in that, described background calculus of differences adopts edge detection algorithm commonly used.
5. many Method of Vehicle Segmentations that shade connects in the video according to claim 4 is characterized in that described edge detection algorithm comprises SOBEL operator and Canny operator.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210525812.9A CN103049738B (en) | 2012-12-07 | 2012-12-07 | Many Method of Vehicle Segmentations that in video, shade connects |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210525812.9A CN103049738B (en) | 2012-12-07 | 2012-12-07 | Many Method of Vehicle Segmentations that in video, shade connects |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103049738A true CN103049738A (en) | 2013-04-17 |
| CN103049738B CN103049738B (en) | 2016-01-20 |
Family
ID=48062371
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210525812.9A Expired - Fee Related CN103049738B (en) | 2012-12-07 | 2012-12-07 | Many Method of Vehicle Segmentations that in video, shade connects |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103049738B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104125386A (en) * | 2013-04-25 | 2014-10-29 | 宏达国际电子股份有限公司 | Image processing apparatus and image processing method |
| CN104598906A (en) * | 2015-03-04 | 2015-05-06 | 西安工业大学 | External-profile detection method and device for vehicles |
| CN107341810A (en) * | 2017-06-16 | 2017-11-10 | 重庆交通大学 | A kind of automatic vehicle identification method, apparatus and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101030256A (en) * | 2006-02-28 | 2007-09-05 | 沈阳东软软件股份有限公司 | Method and apparatus for cutting vehicle image |
| CN101729872A (en) * | 2009-12-11 | 2010-06-09 | 南京城际在线信息技术有限公司 | Video monitoring image based method for automatically distinguishing traffic states of roads |
-
2012
- 2012-12-07 CN CN201210525812.9A patent/CN103049738B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101030256A (en) * | 2006-02-28 | 2007-09-05 | 沈阳东软软件股份有限公司 | Method and apparatus for cutting vehicle image |
| CN101729872A (en) * | 2009-12-11 | 2010-06-09 | 南京城际在线信息技术有限公司 | Video monitoring image based method for automatically distinguishing traffic states of roads |
Non-Patent Citations (2)
| Title |
|---|
| GOO JUN ET AL: "Tracking and Segmentation of Highway Vehicles in Cluttered and Crowded Scenes", 《IEEE 2008 WORKSHOPS ON APPLICATIONS OF COMPUTER VISION》 * |
| 王菁菁，罗立民: "一种基于区域的车辆阴影消除方法", 《电脑知识与技术》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104125386A (en) * | 2013-04-25 | 2014-10-29 | 宏达国际电子股份有限公司 | Image processing apparatus and image processing method |
| CN104125386B (en) * | 2013-04-25 | 2018-04-17 | 宏达国际电子股份有限公司 | Image processor and its image treatment method |
| CN104598906A (en) * | 2015-03-04 | 2015-05-06 | 西安工业大学 | External-profile detection method and device for vehicles |
| CN104598906B (en) * | 2015-03-04 | 2017-08-22 | 西安工业大学 | Vehicle outline detection method and its device |
| CN107341810A (en) * | 2017-06-16 | 2017-11-10 | 重庆交通大学 | A kind of automatic vehicle identification method, apparatus and electronic equipment |
| CN107341810B (en) * | 2017-06-16 | 2020-07-10 | 重庆交通大学 | Vehicle automatic identification method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103049738B (en) | 2016-01-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104392468B (en) | Based on the moving target detecting method for improving visual background extraction | |
| Li et al. | Real-time moving vehicle detection, tracking, and counting system implemented with OpenCV | |
| WO2022027931A1 (en) | Video image-based foreground detection method for vehicle in motion | |
| CN105427626A (en) | Vehicle flow statistics method based on video analysis | |
| CN110415268A (en) | A kind of moving region foreground image algorithm combined based on background differential technique and frame difference method | |
| Lin et al. | An efficient and robust moving shadow removal algorithm and its applications in ITS | |
| CN104376580B (en) | The processing method of non-interesting zone issue in a kind of video frequency abstract | |
| CN103049738B (en) | Many Method of Vehicle Segmentations that in video, shade connects | |
| Fung et al. | Effective moving cast shadow detection for monocular color image sequences | |
| CN103065145A (en) | Vehicle movement shadow eliminating method | |
| CN103793921A (en) | Moving object extraction method and moving object extraction device | |
| CN105139358A (en) | Video raindrop removing method and system based on combination of morphology and fuzzy C clustering | |
| Aung et al. | Study for license plate detection | |
| Chin et al. | Extended fuzzy background modeling for moving vehicle detection using infrared vision | |
| CN111724416A (en) | Moving object detection method and trajectory tracking method based on background subtraction | |
| CN111724415A (en) | Video image-based multi-target motion detection and tracking method in fixed scene | |
| Hu et al. | An improved vehicle detection and tracking model | |
| Turkane et al. | Moving object detection in video surveillance system | |
| CN105335782A (en) | Image-based target object counting method and apparatus | |
| Pham et al. | A hybrid shadow removal algorithm for vehicle classification in traffic surveillance system | |
| Liang et al. | Real-time Background Subtraction under Varying Lighting Conditions | |
| Lakhotiya et al. | Human object tracking using background subtraction and shadow removal technique | |
| Xua et al. | A novel algorithm of moving cast shadow suppression | |
| As' ari et al. | Object's shadow removal with removal validation | |
| Al-Salih et al. | Object detection and features extraction in video frames using direct thresholding |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160120 Termination date: 20161207 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |