CN103049788B - Based on space number for the treatment of object detection system and the method for computer vision - Google Patents
Based on space number for the treatment of object detection system and the method for computer vision Download PDFInfo
- Publication number
- CN103049788B CN103049788B CN201210567978.7A CN201210567978A CN103049788B CN 103049788 B CN103049788 B CN 103049788B CN 201210567978 A CN201210567978 A CN 201210567978A CN 103049788 B CN103049788 B CN 103049788B
- Authority
- CN
- China
- Prior art keywords
- region
- pedestrian
- width
- image
- people
- 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.)
- Expired - Fee Related
Links
Abstract
The present invention relates to a kind of crossing based on computer vision and treat space number object detection system and method, system comprises two frame ccd video camera and Computerized image processing systems, Computerized image processing system is connected with video camera by interface, two frame ccd video cameras are installed on the two ends of crossing, and the field angle of two frame ccd video cameras comprises the wait pedestrian region on opposite and the vehicle region at opposite slightly to the right or left; Detection method is obtain the image in video image, to the pre-service of pedestrian's waiting area, obtain foreground picture by Gaussian mixture model-universal background model method again, and obtain vertical integral projection figure, it to be processed and Information Statistics finally obtain the number that crossing treats by pedestrian.The present invention compares existing pedestrian's number detection method and has simple advantage efficiently, and method also can be used in complicated case.
Description
Technical field
The present invention relates to technical field of computer vision, space number object detection method treated by the crossing specifically based on computer vision.
Background technology
At present, in intelligent transportation and computer vision field, the determination and analysis of pedestrian is a part and parcel.The examination and analysb technology of pedestrian have studied more than ten years, but still neither one standard, accurately, high performance, and real-time pedestrian detection and analytical algorithm.Due to some characteristics that pedestrian is intrinsic, the complicacy of application scenarios, influencing each other between person to person or human and environment, makes the determination and analysis of pedestrian be a challenge the most difficult in computer vision research field.
In more than ten years in the past, many existing detection methods have been created when pedestrian detection technology obtains academia and engineering circles pays close attention to widely and studies, as the people such as HaritaogluGavrila utilize the contour feature of human body to detect pedestrian, due to human body in the body axle line coordinates presenting certain symmetry, therefore, the projection histogram of certain region internal object profile in horizontal and vertical directions can be calculated, analyze symmetry, to determine that whether target is for pedestrian.Target after a motion segmentation ellipse mates by the people such as Rivlin, Senior, oval major and minor axis and length ratio thereof, and the angle that major and minor axis is formed between plane of delineation coordinate system can be classified to pedestrian as shape facility.The people such as Lipton define moving target rim circumference square with area ratio as dispersion, utilize this feature to distinguish the object such as pedestrian, automobile.The people such as Collins have merged above multiple parameter, use the area, length breadth ratio, dispersion etc. of target as feature, trained a three-layer neural network and classify to targets such as pedestrian, vehicle and crowds.All there are some defects in these four methods above, first, than being easier to the impact being subject to noise, the action change of pedestrian, the complexity of background, all can the extraction of destructive characteristics.Secondly, for shape facility, owing to being analyze the foreground area of segmentation, therefore, they extremely rely on the performance of dispenser, and background segmentation techniques still exists many problems and needs to solve.
Summary of the invention
Method for the existing detection number of people is not suitable for the problem of the pedestrian detection in complex environment, the invention provides a kind of crossing and treat space number object detection method, the number detecting pedestrian that not only can be more accurate, and it is simply effective, goes for comparatively complex environment.
The present invention is achieved by the following technical solutions:
Based on space number of the treating object detection system of computer vision, comprise two frame ccd video camera and Computerized image processing systems; Described Computerized image processing system is connected with video camera by interface; Described two frame ccd video cameras are installed on the two ends of crossing; The field angle of described two frame ccd video cameras comprises the wait pedestrian region on opposite and the vehicle region at opposite slightly to the right or left.
Based on the detection method of space number of the treating object detection system of computer vision, comprise the following steps:
1) video camera capture video images;
2) image in video image is carried out to segmentation and the light intensity adaptive change pre-service of pedestrian's waiting area;
3) employing realizes context update based on Gaussian mixture model-universal background model method, and obtains the foreground picture of pedestrian's waiting area and carry out pre-service to it;
4) method based on vertical integral projection processes to pretreated foreground picture the vertical integral projection figure obtaining pedestrian's waiting area;
5) by finally obtaining the number that crossing is treated by pedestrian to the process of the vertical integral projection figure of pedestrian's waiting area and Information Statistics.
Abovementioned steps 2) in light intensity adaptive change pre-service comprise,
2-1) coloured image of every two field picture of camera acquisition is converted into gray level image;
2-2) add up according to grey level histogram, judge the intensity of light and the contrast of image in image;
Getting grey level statistics is
Wherein, H
ifor the probability that each gray level occurs, i is gray level progression;
If 2-3) L >=O.8 or L≤0.2, carry out setting contrast to current frame image.
Abovementioned steps 3) in choose when pedestrian begins through lateral road and start to upgrade background, select to extract foreground picture during 1s ~ 2s before vehicle pass-through terminates.
Abovementioned steps 3) in the pre-service of foreground picture for carry out histogram equalization successively to foreground image, medium filtering, connected domain denoising, expansive working.
Abovementioned steps 4) in, vertical integral projection V is,
Wherein P (i, j) represent the pixel value that foreground picture is corresponding with i position be 0 pixel count, W is the width of pretreated image, and H is the height of pretreated image.
Abovementioned steps 5) comprise,
5-1) delimit threshold line, the line point being greater than threshold value is the vertical region in the potential region of the number of people, and removes the point in the potential region of the non-number of people;
The threshold value of 5-2) delimiting the potential region of the number of people removes noise spot;
5-3) to the width in effective vertically region with highly add up, determine width shared by a people and highly;
5-4) by every block effectively vertically the width in region and height with obtain this time detect in width shared by a people and highly comparing, determine number contained by this region;
5-5) quantity of the pedestrian in the every block counted effectively vertically region is all added up, as this detect the final wait obtained the quantity of pedestrian of lateral road.
Abovementioned steps 5-1) in delimit threshold line be look for according to the distribution situation of the integral projection of vertical direction is adaptive
To a rational threshold value Threshold,
Wherein P
ifor V
(i), the probability of i ∈ [1, W], α is scale factor, α ∈ [0.6,1].
Abovementioned steps 5-2) in threshold value be 20 pixels.
Abovementioned steps 5-3) refer to: to the width in effective vertically region with highly add up, first obtain minimum width and height, again on this basis, the width in being determined this time to detect by the mean value differing the value being less than 5 pixels with minimum value shared by a people and highly.
Abovementioned steps 5-4) refer to:
If one piece of vertical region is more than the width shared by a people with highly all in threshold value allowed band, this region only comprises a pedestrian;
When one piece of vertical region more than the value of the width shared by a people and height when threshold value allowed band is outer, only width then comprises two pedestrians in this region more than the width shared by a people, only highly then comprise two pedestrians in this region more than the height shared by a people, if width and highly all more than the width shared by a people and highly, this region comprises four pedestrians.
The threshold value of aforesaid threshold values allowed band is arranged according to the position of the concrete condition at crossing and video camera erection.
The height threshold of aforesaid threshold values is 12 pixels, and width threshold value is 17 pixels.
Advantage of the present invention is, method is simply effective, the detection people counting method comparing feature based coupling is easily subject to surrounding objects interference makes the situation that error under complex situations is larger, the present invention is more suitable for operating in complex situations, the intelligent traffic light supervisory systems applying to crossing place is particularly applicable, because principle is simple, also can be suitable reduce costs.
Accompanying drawing explanation
Fig. 1 is that space number object detection method process flow diagram treated by the crossing that the present invention is based on computer vision;
Fig. 2 is that the present invention is to the process of vertical integral projection figure and Information Statistics analysis process figure.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is further described.
Based on space number of the treating object detection system of computer vision, comprise two frame ccd video camera and Computerized image processing systems; Described Computerized image processing system is connected with video camera by interface; Described two frame ccd video cameras are installed on the two ends of crossing; The field angle of described two frame ccd video cameras comprises the wait pedestrian region on opposite and the vehicle region at opposite slightly to the right or left.
As shown in Figure 1, the detection method of space number object detection system treated by the crossing based on computer vision, comprises the following steps:
1) video camera capture video images;
2) image in video image is carried out to segmentation and the light intensity adaptive change pre-service of pedestrian's waiting area;
3) employing obtains the foreground picture of pedestrian's waiting area based on Gaussian mixture model-universal background model method and carries out pre-service to it;
4) method based on vertical integral projection processes to pretreated foreground picture the vertical integral projection figure obtaining pedestrian's waiting area;
5) by finally obtaining the number that crossing is treated by pedestrian to the process of the vertical integral projection figure of pedestrian's waiting area and Information Statistics.
Specifically be divided into following four parts:
Part I, get the pre-service such as segmentation and light intensity adaptive change that the image obtained in video image carries out pedestrian's waiting area, specifically comprise the following steps:
(1) segmentation of pedestrian standing area.The maximum feature of image to provide a large amount of information, meanwhile also can bring a large amount of interfere informations.The object being partitioned into pedestrian standing area from image is that the effective coverage of pedestrian being stood extracts, and reduces part interfere information to a certain extent simultaneously.
(2) light intensity adaptive change.For concrete crossing, the light intensity caused due to reasons such as weather is different, brings very large inconvenience can to the detection of pedestrian.When light in sequence of video images very weak or very strong time, target and background has very large similarity, and carry out in the gray-scale map that frame-to-frame differences computing obtains, target is not easy to be detected, and the impact of noise is also relatively large.So after the standing area being partitioned into pedestrian, first to ccd video camera gather every two field picture (picture traverse is W, highly for H) coloured image be converted into gray level image, then add up according to grey level histogram, judge the power of light and the contrast of image in image, the image more weak or stronger to light carries out process based on pixel in order to reach the object of image enhaucament, concrete deterministic process is as follows: if present frame gray picture, I (x, y), the probability H that each gray level occurs is added up
i, grey level statistics is taken as:
if L>=O.8 or L≤0.2, carry out setting contrast to current frame image.
Part II, adopts and obtains the foreground picture of pedestrian's waiting area based on Gaussian mixture model-universal background model method and carry out pre-service to it, specifically comprise the following steps:
(1) real-time update of background.The method of mixture Gaussian background model is adopted to realize the real-time update of background, added up by the frequency occurred the corresponding grey scale value of each pixel in continuous multiple frames image, when detected target is in the process of motion, in all gray-scale values that same pixel place occurs, the number of times that the gray-scale value of the pixel in background image occurs wherein is maximum, the frequency that the gray-scale value of the pixel namely in background image occurs is higher than the frequency of occurrences of other gray-scale values, according to this principle, after statistics terminates, the gray-scale value of gray-scale value the highest for the frequency of occurrences as corresponding pixel points place is saved, restore view picture background image again, finally preserve this background image for future use.For the opportunity of pedestrian standing area context update, through lot of experiments, final selection starts when pedestrian begins through lateral road to upgrade background (can suitably postpone a bit of time), and the closing time of context update should the pedestrian a little earlier in setting pass through the time (because in last 1 to 2 seconds, pedestrian may stop wait for next time pass through, if continue to upgrade background, then the pedestrian now stopped also may be regarded background).
(2) acquisition of the prospect of detected object is comprised.The very last seconds clock that will terminate at vehicle pass-through is selected in the present invention, during as 1s ~ 2s, obtain the two field picture comprising pedestrian, the quantity information relatively truth (time that the time terminates the closer to vehicle pass-through, the information comprised in this width image is more close to truth) of the pedestrian comprised in this two field picture.Again this two field picture obtained and the background image obtained are done frame difference above, thus obtain foreground image.
(3) to the pre-service of foreground image.In order to further reduce interference, to obtain foreground image carried out pre-service, processing procedure for carry out histogram equalization successively to foreground image, medium filtering, connected domain denoising, expansive working.
Part III, the method based on vertical integral projection processes to the foreground picture after denoising the vertical integral projection figure obtaining pedestrian's waiting area.If pre-service obtains the binary map G that size is W*H, W is the width of pretreated image, H is the height of pretreated image, notice the equal vanishing of background and partial target region at this moment, and in binary map, the part that on vertical direction, pixel number is more, substantially from the vertical area part at the head place of human body, based on such analysis, determines the vertical region at number of people place according to the integral projection of figure G vertical direction.Note V represents the integral projection of the vertical direction of figure G, and V is a W dimensional vector, wherein
P (i, j) represent pixel value that foreground image is corresponding with i position be 0 pixel count, W is picture traverse, and H is picture altitude.
Part IV, finally obtains by the process of the vertical integral projection figure to pedestrian's waiting area and Information Statistics the number that crossing treats by pedestrian, as shown in Figure 2, specifically comprises the following steps:
(1) delimit threshold line, determine the vertical region in the potential region of the number of people, and remove the point in the potential region of the non-number of people.In order to determine the vertical region at potential number of people place, the present invention finds a rational threshold value to mark off these regions according to the distribution situation of the integral projection of figure G vertical direction is adaptive, and method is: statistics V
(i), the probability distribution of i ∈ [1, W], note V
(i), the probability of i ∈ [1, W] is P
i, threshold value is taken as:
an i.e. average statistical part of threshold value amount of orientation V, wherein α is a scale factor, α ∈ [0.6,1], and its value is relevant to the brightness gathering image, is calculated by the grey level histogram adding up the present image obtained.By a large amount of tests, L≤0.6, α gets 0.6, and in other situations, α gets the value of L, and effectiveness comparison is good.
(2) delimit threshold value vertical region too little for width is removed as noise spot.After obtaining the vertical region at potential number of people place, through a large amount of tests, choose suitable threshold value, vertical region too little for width is removed, in the present embodiment, threshold value elects 20 pixels as, and this threshold value needs to measure according to the crossing situation of reality and the position of video camera erection.
(3) to the width in effective vertically region with highly add up, width shared by a people and is highly determined.To the width in effective vertically region with highly add up, first obtain minimum width and height, again on this basis, the width in the mean value decision this time detection of several values of be more or less the same with minimum value (being less than 5 pixels as differed) shared by a people and height is obtained.
(4) by every block effectively vertically the width in region and height with obtain this time detect in width shared by a people and highly comparing, determine number contained by this region.If time relatively when the width in some regions or highly exceed acquisition this time detection in width shared by a people or highly certain threshold value time, then think do not only have a pedestrian in this vertical region, the final number in this region is determined according to the difference of threshold value, the setting of this threshold value needs to measure according to the crossing situation of reality and the position of video camera erection, in the present embodiment, height threshold is 12 pixels, width threshold value is 17 pixels, the method that contained by the vertical region of the every block of concrete statistics, number is how many only comprises a pedestrian more than the width shared by a people with highly all still thinking that in threshold value allowed band this region be can be regarded as: one piece of vertical region, if and one piece of vertical region more than the value of the width shared by a people and height when threshold value allowed band is outer, only more than the width shared by a people, width then thinks that this region comprises two pedestrians, only highly then think that this region comprises two pedestrians more than the height shared by a people, if width and highly all more than the width shared by a people and highly, think that this region comprises four pedestrians.
(5) quantity of the pedestrian in the every block counted effectively vertically region is all added up, as this detect the final wait obtained the quantity of pedestrian of lateral road.
Claims (11)
1., based on the detection method of space number of the treating object detection system of computer vision, it is characterized in that: comprise the following steps:
1) set up space number of the treating object detection system based on computer vision, comprise two frame ccd video camera and Computerized image processing systems; Described Computerized image processing system is connected with video camera by interface; Described two frame ccd video cameras are installed on the two ends of crossing; The field angle of described two frame ccd video cameras comprises the wait pedestrian region on opposite and the vehicle region at opposite slightly to the right or left;
2) video camera capture video images;
3) image in video image is carried out to segmentation and the light intensity adaptive change pre-service of pedestrian's waiting area;
4) employing realizes context update based on Gaussian mixture model-universal background model method, and obtains the foreground picture of pedestrian's waiting area and carry out pre-service to it;
5) method based on vertical integral projection processes to pretreated foreground picture the vertical integral projection figure obtaining pedestrian's waiting area;
6) by finally obtaining the number that crossing is treated by pedestrian to the process of the vertical integral projection figure of pedestrian's waiting area and Information Statistics;
6-1) delimit threshold line, the line point being greater than threshold value is the vertical region in the potential region of the number of people, and removes the point in the potential region of the non-number of people;
The threshold value of 6-2) delimiting the potential region of the number of people removes noise spot;
6-3) to the width in effective vertically region with highly add up, determine width shared by a people and highly;
6-4) by every block effectively vertically the width in region and height with obtain this time detect in width shared by a people and highly comparing, determine number contained by this region;
6-5) quantity of the pedestrian in the every block counted effectively vertically region is all added up, as this detect the final wait obtained the quantity of pedestrian of lateral road.
2. detection method according to claim 1, is characterized in that: described step 3) in light intensity adaptive change pre-service comprise,
3-1) coloured image of every two field picture of camera acquisition is converted into gray level image;
3-2) add up according to grey level histogram, judge the intensity of light and the contrast of image in image; Getting grey level statistics is
Wherein, H
ifor the probability that each gray level occurs, i is gray level progression;
If 3-3) L >=0.8 or L≤0.2, carries out setting contrast to current frame image.
3. detection method according to claim 1, is characterized in that: described step 4) in, choose when pedestrian begins through lateral road and start to upgrade background, select to extract foreground picture during 1s ~ 2s before vehicle pass-through terminates.
4. detection method according to claim 1, is characterized in that: described step 4) in the pre-service of foreground picture for carry out histogram equalization successively to foreground image, medium filtering, connected domain denoising, expansive working.
5. detection method according to claim 1, is characterized in that: described step 5) in, vertical integral projection V is,
Wherein P (i, j) represent the pixel value that foreground picture is corresponding with i position be not 0 pixel count, W is the width of pretreated image, and H is the height of pretreated image.
6. detection method according to claim 5, is characterized in that: described step 6-1) in delimit threshold line be find a rational threshold value Threshold according to the distribution situation of the integral projection of vertical direction is adaptive,
Wherein P
ifor V
(i), the probability of i ∈ [1, W], α is scale factor, α ∈ [0.6,1].
7. detection method according to claim 1, is characterized in that: described step 6-2) in threshold value be 20 pixels.
8. detection method according to claim 1, it is characterized in that: described step 6-3) refer to: to the width in effective vertically region with highly add up, first obtain minimum width and height, again on this basis, the width in being determined this time to detect by the mean value differing the value being less than 5 pixels with minimum value shared by a people and highly.
9. detection method according to claim 1, is characterized in that: described step 6-4) refer to:
If one piece of vertical region is more than the width shared by a people with highly all in threshold value allowed band, this region only comprises a pedestrian;
When one piece of vertical region more than the value of the width shared by a people and height when threshold value allowed band is outer, only width then comprises two pedestrians in this region more than the width shared by a people, only highly then comprise two pedestrians in this region more than the height shared by a people, if width and highly all more than the width shared by a people and highly, this region comprises four pedestrians.
10. detection method according to claim 9, is characterized in that: described threshold value is arranged according to the position of the concrete condition at crossing and video camera erection.
11. detection methods according to claim 10, is characterized in that: the height threshold of described threshold value is 12 pixels, and width threshold value is 17 pixels.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210567978.7A CN103049788B (en) | 2012-12-24 | 2012-12-24 | Based on space number for the treatment of object detection system and the method for computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210567978.7A CN103049788B (en) | 2012-12-24 | 2012-12-24 | Based on space number for the treatment of object detection system and the method for computer vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103049788A CN103049788A (en) | 2013-04-17 |
| CN103049788B true CN103049788B (en) | 2015-11-18 |
Family
ID=48062420
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210567978.7A Expired - Fee Related CN103049788B (en) | 2012-12-24 | 2012-12-24 | Based on space number for the treatment of object detection system and the method for computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103049788B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105335985B (en) * | 2014-08-01 | 2019-03-01 | 深圳中集天达空港设备有限公司 | A kind of real-time capturing method and system of docking aircraft based on machine vision |
| CN105373135B (en) * | 2014-08-01 | 2019-01-01 | 深圳中集天达空港设备有限公司 | A kind of method and system of aircraft docking guidance and plane type recognition based on machine vision |
| CN105302151B (en) * | 2014-08-01 | 2018-07-13 | 深圳中集天达空港设备有限公司 | A kind of system and method for aircraft docking guiding and plane type recognition |
| CN105744224A (en) * | 2016-02-04 | 2016-07-06 | 张宏业 | Suspended type high and low voltage power distribution cabinet |
| CN105893962A (en) * | 2016-03-31 | 2016-08-24 | 成都信息工程大学 | Method for counting passenger flow at airport security check counter |
| CN108804992B (en) * | 2017-05-08 | 2022-08-26 | 电子科技大学 | Crowd counting method based on deep learning |
| CN107133610B (en) * | 2017-06-01 | 2020-09-01 | 电子科技大学 | Visual detection and counting method for traffic flow under complex road conditions |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101799987A (en) * | 2010-03-10 | 2010-08-11 | 北京航空航天大学 | Self-adaptive intelligent traffic light and control method thereof |
| CN101847206A (en) * | 2010-04-21 | 2010-09-29 | 北京交通大学 | Pedestrian traffic statistical method and system based on traffic monitoring facilities |
| US8331729B2 (en) * | 2008-01-16 | 2012-12-11 | Samsung Electronics Co., Ltd. | Block-based object counting apparatus and method |
| CN102831592A (en) * | 2012-08-10 | 2012-12-19 | 中国电子科技集团公司第四十一研究所 | Image nonlinearity enhancement method based on histogram subsection transformation |
-
2012
- 2012-12-24 CN CN201210567978.7A patent/CN103049788B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8331729B2 (en) * | 2008-01-16 | 2012-12-11 | Samsung Electronics Co., Ltd. | Block-based object counting apparatus and method |
| CN101799987A (en) * | 2010-03-10 | 2010-08-11 | 北京航空航天大学 | Self-adaptive intelligent traffic light and control method thereof |
| CN101847206A (en) * | 2010-04-21 | 2010-09-29 | 北京交通大学 | Pedestrian traffic statistical method and system based on traffic monitoring facilities |
| CN102831592A (en) * | 2012-08-10 | 2012-12-19 | 中国电子科技集团公司第四十一研究所 | Image nonlinearity enhancement method based on histogram subsection transformation |
Non-Patent Citations (1)
| Title |
|---|
| 基于Hough 变换头部检测与跟踪的方法研究;卢湖川 等;《系统仿真学报》;20080430;第20卷(第8期);第2128页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103049788A (en) | 2013-04-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103049788B (en) | Based on space number for the treatment of object detection system and the method for computer vision | |
| CN103077384B (en) | A kind of method and system of vehicle-logo location identification | |
| Xia et al. | Towards improving quality of video-based vehicle counting method for traffic flow estimation | |
| CN102509291B (en) | Pavement disease detecting and recognizing method based on wireless online video sensor | |
| Lai et al. | Image-based vehicle tracking and classification on the highway | |
| CN102542289B (en) | Pedestrian volume statistical method based on plurality of Gaussian counting models | |
| CN108596129A (en) | A kind of vehicle based on intelligent video analysis technology gets over line detecting method | |
| CN104036262B (en) | A kind of method and system of LPR car plates screening identification | |
| CN101872546A (en) | Video-based method for rapidly detecting transit vehicles | |
| Guo et al. | Image-based seat belt detection | |
| CN106022243B (en) | A kind of retrograde recognition methods of the car lane vehicle based on image procossing | |
| CN103824070A (en) | Rapid pedestrian detection method based on computer vision | |
| CN108921076B (en) | Pavement crack disease self-adaptive constant false alarm detection method based on image | |
| CN105321189A (en) | Complex environment target tracking method based on continuous adaptive mean shift multi-feature fusion | |
| CN111505632A (en) | Ultra-wideband radar action attitude identification method based on power spectrum and Doppler characteristics | |
| CN107516423B (en) | Video-based vehicle driving direction detection method | |
| CN107578048A (en) | A kind of long sight scene vehicle checking method based on vehicle rough sort | |
| CN104463104A (en) | Fast detecting method and device for static vehicle target | |
| FAN et al. | Robust lane detection and tracking based on machine vision | |
| Helala et al. | Road boundary detection in challenging scenarios | |
| CN108921147A (en) | A kind of black smoke vehicle recognition methods based on dynamic texture and transform domain space-time characteristic | |
| Zhang et al. | A front vehicle detection algorithm for intelligent vehicle based on improved gabor filter and SVM | |
| Chen et al. | Block-constraint line scanning method for lane detection | |
| Chen et al. | Automatic head detection for passenger flow analysis in bus surveillance videos | |
| Płaczek | A real time vehicle detection algorithm for vision-based sensors |
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: 20151118 Termination date: 20161224 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |