CN105279754A - Part segmentation method suitable for bicycle video detection - Google Patents
Part segmentation method suitable for bicycle video detection Download PDFInfo
- Publication number
- CN105279754A CN105279754A CN201510575140.6A CN201510575140A CN105279754A CN 105279754 A CN105279754 A CN 105279754A CN 201510575140 A CN201510575140 A CN 201510575140A CN 105279754 A CN105279754 A CN 105279754A
- Authority
- CN
- China
- Prior art keywords
- bicycle
- sample
- picture
- parts
- database
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Abstract
The invention discloses a part segmentation method suitable for bicycle video detection. The method includes the following steps of: (1) acquiring the basic video data of a target camera; (2) building a bicycle sample database; (3) performing gray-scale transformation on samples of the database; (4) acquiring a bicycle sample average image; (5) performing part partition on the bicycle sample average image; and (6) determining part segmentation proportion. The part segmentation method suitable for bicycle video detection provides specific part segmentation basis for the bicycle detection algorithm of different cameras, improves the accuracy of the bicycle video detection algorithm, and has actual popularization values.
Description
Technical field
The present invention relates to traffic video detection field, particularly a kind of parts dividing method being applicable to bicycle video and detecting.
Background technology
Video bicycle detection technique refers to and identify bicycle in monitor video.Along with the development of road traffic video capture technology, a large amount of monitoring cameras is applied in crossing and access etc.Computing machine is utilized to arise at the historic moment to the demand that the automobile in monitor video, bicycle, pedestrian etc. detect.Current parts are segmented in apply in Video Detection Algorithm many, better can improve recognition efficiency, but the parts dividing method not yet having the bicycle sample of corresponding method to target video camera to carry out effectively and reasonably.
Current bicycle Video Detection Algorithm is not yet applied to parts dividing method, not yet finds the article being applicable to the parts dividing method that bicycle detects for target video camera and patent description.
Summary of the invention
In order to overcome the shortcoming of prior art existence with not enough, the invention provides a kind of parts dividing method being applicable to bicycle video and detecting.
The present invention adopts following technical scheme:
Be applicable to the parts dividing method that bicycle video detects, comprise the steps:
S1 obtains the video data of target video camera;
S2 builds bicycle sample database, is specially: extract dynamically self car picture in video data, be then normalized, and the bicycle picture after process builds bicycle sample database;
S3 carries out gradation conversion to the picture of bicycle sample database;
Sample database picture after gradation conversion is carried out overlap-add procedure by S4, obtains bicycle sample mean figure;
S5 carries out parts division to bicycle sample mean figure, specifically divides: head, health and bicycle body three part;
S6 determines the parts ration of division of target video camera, is specially: the ratio obtaining its mean pixel according to the head of bicycle sample mean figure, health and bicycle body, is the parts ration of division of target video camera.
Carrying out gradation conversion in described S3 utilizes the cross-platform computer vision library OpenCV increased income to carry out greyscale transformation to the samples pictures in database.
Described target video camera is at least one, at least comprises bicycle sample size and be greater than 1000 in described video.
Extract dynamically self car picture in video data in S2 and specifically adopt background subtraction, then screen qualified bicycle picture as sample, finally adopt the Resize () function in OpenCV by unified for the bicycle picture sample for 64*128 pixel size.
The overlap-add procedure of described S4 is specifically by sample gray matrix M in database
iaverage, obtain gray average matrix M, and M is inputted the imshow () function in OpenCV, thus obtain bicycle sample mean figure.
Beneficial effect of the present invention:
The present invention analyzes the bicycle samples pictures that single or multiple video camera is taken; Method is simple, and application the method takes into full account the sample of shot by camera and the impact of environment, and can improve efficiency and the progress of bicycle detection algorithm, the present invention has very large actual promotional value.
Accompanying drawing explanation
Fig. 1 is the step frame diagram of one embodiment of the present of invention;
Fig. 2 is the process flow diagram of the acquisition dynamic object picture of one embodiment of the present of invention;
Fig. 3 is the good sample instance graph of one embodiment of the present of invention;
Fig. 4 is the process flow diagram of the acquisition bicycle sample mean figure of one embodiment of the present of invention;
Fig. 5 is the bicycle sample mean figure of one embodiment of the present of invention;
Fig. 6 is that the bicycle sample mean figure parts of one embodiment of the present of invention divide schematic diagram.
Embodiment
Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited thereto.
Embodiment
As shown in figs 1 to 6, a kind of parts dividing method being applicable to bicycle video and detecting, comprises the steps:
S1 obtains the video data of target video camera, obtains the video data in 12 hours of three video cameras in the present embodiment;
S2 builds bicycle sample database, is specially and extracts dynamically self car picture in video data, be then normalized, and the bicycle picture after process builds bicycle sample database, and described dynamically self car picture is specifically people's picture by bike;
It is adopt background subtraction that the present invention extracts dynamically self car picture, and described background subtraction is specially step and is:
S2.1 carries out the pre-service of image, mainly comprises and carries out gray processing and filtering to image;
S2.2 background modeling
Carry out interval statistics according to the gray-scale value of front N two field picture thus obtain the initial background that has statistical significance, N=10 here.
S2.3 foreground extraction:
Current up-to-date image and background are done difference, can try to achieve background subtraction figure, then carry out binaryzation to this figure, final acquisition sport foreground region, namely realizes Iamge Segmentation.
To the dynamic object picture extracted, reject as vehicle, bicycle, bicycle shade etc., filter out preferably, have the picture of bicycle whole body as sample, the sample of same bicycle is no more than 20 pictures, as shown in Figure 3.
Adopt the Resize () function in OpenCV by unified for the bicycle samples pictures of the different size sample for 64*128 pixel size, form bicycle sample database.
S3 carries out gradation conversion to the picture of bicycle sample database as shown in Figure 4;
Then utilize the Cvtcolor () function in OpenCV, carry out greyscale transformation to all samples pictures in database, each like this sample obtains a corresponding gray matrix M
i;
Then by sample gray matrix M all in database
iaverage, obtain sample gray average matrix M, formula is as follows:
S4 the imshow () function inputted by M in OpenCV show, thus obtain bicycle sample mean figure.
S5 is to three parts of bicycle sample mean figure, and described three parts comprise head, health and bicycle body.The division of described three parts can adopt artificial division method or software demarcation method.
S6 determines that the pixel of three parts is high, and averages respectively, obtains H
head, H
bodyand H
bike, and then calculate average proportions H
head: H
body: H
bike, result is as the parts ration of division of target video camera.
In the present embodiment, Fig. 2 is the picture adopting background subtraction to obtain dynamic object, and Fig. 3 is the instance graph of sample, Fig. 5 is that in the present embodiment, bicycle sample mean figure, Fig. 6 are segmentation results of the present invention, and the pixel then obtaining different parts is high, and average respectively, obtain:
H
head=15.2 (pixels), H
body=41.4 (pixels), H
bike=71.4 (pixels)
Finally calculating average proportions 1:2.72:4.70 again, is exactly the parts ration of division of target video camera.
The present invention carries out Treatment Analysis to the bicycle samples pictures that single or multiple video camera is taken, and is obtained from driving sample mean figure, finally determines the bicycle sample components ration of division.
Above-described embodiment is the present invention's preferably embodiment; but embodiments of the present invention are not limited by the examples; change, the modification done under other any does not deviate from Spirit Essence of the present invention and principle, substitute, combine, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.
Claims (5)
1. be applicable to the parts dividing method that bicycle video detects, it is characterized in that, comprise the steps:
S1 obtains the video data of target video camera;
S2 builds bicycle sample database, is specially: extract dynamically self car picture in video data, be then normalized, and the bicycle picture after process builds bicycle sample database;
S3 carries out gradation conversion to the picture of bicycle sample database;
Sample database picture after gradation conversion is carried out overlap-add procedure by S4, obtains bicycle sample mean figure;
S5 carries out parts division to bicycle sample mean figure, specifically divides: head, health and bicycle body three part;
S6 determines the parts ration of division of target video camera, is specially: the ratio obtaining its mean pixel according to the head of bicycle sample mean figure, health and bicycle body, is the parts ration of division of target video camera.
2. parts dividing method according to claim 1, is characterized in that, carries out gradation conversion and utilize the cross-platform computer vision library OpenCV increased income to carry out greyscale transformation to the samples pictures in database in described S3.
3. parts dividing method according to claim 1, is characterized in that, described target video camera is at least one, at least comprises bicycle sample size and be greater than 1000 in described video.
4. parts dividing method according to claim 1, it is characterized in that, extract dynamically self car picture in video data in S2 and specifically adopt background subtraction, then screen qualified bicycle picture as sample, finally adopt the Resize () function in OpenCV by unified for the bicycle picture sample for 64*128 pixel size.
5. parts dividing method according to claim 1, is characterized in that, the overlap-add procedure of described S4 is specifically by sample gray matrix M in database
iaverage, obtain gray average matrix M, and M is inputted the imshow () function in OpenCV, thus obtain bicycle sample mean figure.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510575140.6A CN105279754B (en) | 2015-09-10 | 2015-09-10 | A kind of component dividing method suitable for bicycle video detection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510575140.6A CN105279754B (en) | 2015-09-10 | 2015-09-10 | A kind of component dividing method suitable for bicycle video detection |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105279754A true CN105279754A (en) | 2016-01-27 |
CN105279754B CN105279754B (en) | 2018-06-22 |
Family
ID=55148706
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510575140.6A Active CN105279754B (en) | 2015-09-10 | 2015-09-10 | A kind of component dividing method suitable for bicycle video detection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105279754B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090316957A1 (en) * | 2008-06-23 | 2009-12-24 | Chao-Ho Chen | Method of vehicle segmentation and counting for nighttime video frames |
CN104200210A (en) * | 2014-08-12 | 2014-12-10 | 合肥工业大学 | License plate character segmentation method based on parts |
CN104778453A (en) * | 2015-04-02 | 2015-07-15 | 杭州电子科技大学 | Night pedestrian detection method based on statistical features of infrared pedestrian brightness |
CN104778444A (en) * | 2015-03-10 | 2015-07-15 | 公安部交通管理科学研究所 | Method for analyzing apparent characteristic of vehicle image in road scene |
-
2015
- 2015-09-10 CN CN201510575140.6A patent/CN105279754B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090316957A1 (en) * | 2008-06-23 | 2009-12-24 | Chao-Ho Chen | Method of vehicle segmentation and counting for nighttime video frames |
CN104200210A (en) * | 2014-08-12 | 2014-12-10 | 合肥工业大学 | License plate character segmentation method based on parts |
CN104778444A (en) * | 2015-03-10 | 2015-07-15 | 公安部交通管理科学研究所 | Method for analyzing apparent characteristic of vehicle image in road scene |
CN104778453A (en) * | 2015-04-02 | 2015-07-15 | 杭州电子科技大学 | Night pedestrian detection method based on statistical features of infrared pedestrian brightness |
Also Published As
Publication number | Publication date |
---|---|
CN105279754B (en) | 2018-06-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021208275A1 (en) | Traffic video background modelling method and system | |
CN103258332B (en) | A kind of detection method of the moving target of resisting illumination variation | |
CN102307274B (en) | Motion detection method based on edge detection and frame difference | |
Riche et al. | Rare: A new bottom-up saliency model | |
CN102043950B (en) | Vehicle outline recognition method based on canny operator and marginal point statistic | |
CN112132156A (en) | Multi-depth feature fusion image saliency target detection method and system | |
CN102665034A (en) | Night effect removal method for camera-collected video | |
CN102457724B (en) | Image motion detecting system and method | |
CN103093198A (en) | Crowd density monitoring method and device | |
CN112800860A (en) | Event camera and visual camera cooperative high-speed scattered object detection method and system | |
CN111079669A (en) | Image processing method, device and storage medium | |
CN102610104B (en) | Onboard front vehicle detection method | |
CN104796582A (en) | Video image denoising and enhancing method and device based on random ejection retinex | |
CN115861380B (en) | Method and device for tracking visual target of end-to-end unmanned aerial vehicle under foggy low-illumination scene | |
CN107844772A (en) | A kind of motor vehicle automatic testing method based on movable object tracking | |
CN103295182A (en) | Circuit system and method for performing contrast stretch processing on infrared images | |
CN103971524A (en) | Traffic flow detection method based on machine vision | |
CN104021527A (en) | Rain and snow removal method in image | |
CN110276318A (en) | Nighttime road rains recognition methods, device, computer equipment and storage medium | |
CN107194946B (en) | FPGA-based infrared salient object detection method | |
CN103049788A (en) | Computer-vision-based system and method for detecting number of pedestrians waiting to cross crosswalk | |
CN108009480A (en) | A kind of image human body behavioral value method of feature based identification | |
CN114565597A (en) | Nighttime road pedestrian detection method based on YOLOv3-tiny-DB and transfer learning | |
CN105279754A (en) | Part segmentation method suitable for bicycle video detection | |
CN112446292B (en) | 2D image salient object detection method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |