CN103077380B - A kind of demographic method based on video and device - Google Patents
A kind of demographic method based on video and device Download PDFInfo
- Publication number
- CN103077380B CN103077380B CN201310005046.8A CN201310005046A CN103077380B CN 103077380 B CN103077380 B CN 103077380B CN 201310005046 A CN201310005046 A CN 201310005046A CN 103077380 B CN103077380 B CN 103077380B
- Authority
- CN
- China
- Prior art keywords
- video
- pixel
- pixel point
- frame
- white pixel
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of demographic method based on video and device, solve existing when carrying out demographics, it is impossible to identify the adhesion between person to person's video image, the inaccurate problem of demographics caused.The method includes identifying the sport foreground monitoring video information, and carry out binary conversion treatment for each frame of video, after binary conversion treatment, in each frame of video, sport foreground is white pixel point, background is black pixel point, for each white pixel point, pixel value according to other pixels adjacent with this white pixel point, determine the white pixel being positioned on skeleton point and be positioned at the pixel of head region of people, being tracked statistical number of person according to the pixel of the head region being positioned at people determined.Owing to determining the white pixel point on skeleton in embodiments of the present invention, it is achieved that " thinization " of human body is processed, it is possible to the interpersonal adhesion problems of effective solution, thus improve the accuracy of demographics.
Description
Technical field
The present invention relates to technical field of video image processing, particularly relate to a kind of demographic method based on video and device.
Background technology
Along with socioeconomic development, occurring in that the public places such as increasing amusement, leisure, these public places can be supermarket, chain store, department store, megastore, hotel, KTV etc..How the key monitoring region of the public places such as these amusement, leisure is carried out demographics and analysis, become an important technical problem.
At present, it is possible to carry out demographics by methods such as Intelligent number-of-people statistics, Intelligent number-of-people statistics be video image is analyzed, image procossing, pattern recognition etc. process, to the statistics specifying the number in region to carry out.But this statistical method is primarily directed to indoor or in-plant scene, for outdoor or remote scene, due to distant, between the video image of person to person, it is likely occurred adhesion, and the situation of statistical method None-identified adhesion of the prior art, thus causing the inaccurate of demographics.
Summary of the invention
The embodiment of the present invention provides the demographic method of a kind of video and device, in order to solve in prior art when carrying out demographics, it is impossible to identify the adhesion between person to person's video image, the inaccurate problem of demographics caused.
The embodiment of the present invention provides a kind of demographic method based on video, and described statistical method includes:
Obtain and treat the video information that detection region carries out detecting, identify the sport foreground of each frame of video in this video information；
The each frame of video identified is carried out binary conversion treatment, and after binary conversion treatment, in each frame of video, sport foreground is white pixel point, and background is black pixel point；
For each frame of video after binary conversion treatment, detect each white pixel point in this frame of video, the pixel value according to the pixel adjacent with this white pixel point, it is determined that be positioned at the white pixel point on skeleton；
According to the white pixel being positioned on the skeleton point determined, and the pixel value of adjacent each pixel, it is determined that whether this white pixel point is the pixel of the head region of people；
The pixel determining the head region for people is tracked, according to the result statistical number of person followed the tracks of.
The embodiment of the present invention provides a kind of people counting device based on video, and described statistic device includes:
Identification module, treats, for obtaining, the video information that detection region carries out detecting, identifies the sport foreground of each frame of video in this video information；The each frame of video identified is carried out binary conversion treatment, and after binary conversion treatment, in each frame of video, sport foreground is white pixel point, and background is black pixel point；
First determines module, for for each frame of video after binary conversion treatment, detecting each white pixel point in this frame of video, the pixel value according to the pixel adjacent with this white pixel point, it is determined that be positioned at the white pixel point on skeleton；
Second determines module, for according to the white pixel being positioned on the skeleton point determined, and the pixel value of adjacent each pixel, it is determined that whether this white pixel point is the pixel of the head region of people；
Tracking module, for being tracked the pixel determining the head region for people, according to the result statistical number of person followed the tracks of.
The embodiment of the present invention provides a kind of demographic method based on video and device, the method includes: identify the sport foreground monitoring video information, and carry out binary conversion treatment for each frame of video, after binary conversion treatment, in each frame of video, sport foreground is white pixel point, background is black pixel point, for each white pixel point, pixel value according to other pixels adjacent with this white pixel point, determine the white pixel point being positioned on skeleton, and in these white pixel points, determine the pixel of the head region being positioned at people, pixel according to the head region being positioned at people determined is tracked, according to the result statistical number of person followed the tracks of.Owing to determining the white pixel point on skeleton in embodiments of the present invention, it is achieved that " thinization " of human body is processed, it is possible to the interpersonal adhesion problems of effective solution, thus improve the accuracy of demographics.
Accompanying drawing explanation
The demographics process schematic based on video that Fig. 1 provides for the embodiment of the present invention；
The position relationship schematic diagram of this white pixel point that Fig. 2 provides for the embodiment of the present invention and the adjacent pixel of this white pixel point；
A kind of people counting device structural representation based on video that Fig. 3 provides for the embodiment of the present invention.
Detailed description of the invention
The present invention is in order to carry out demographics for environment such as outdoors, and improves the accuracy of demographics, it is provided that a kind of demographic method based on video and device.
Below in conjunction with Figure of description, the embodiment of the present invention is described in detail.
The demographics process schematic based on video that Fig. 1 provides for the embodiment of the present invention, this process comprises the following steps:
S101: obtain and treat the video information that detection region carries out detecting, identify the sport foreground of each frame of video in this video information.
Photographic head can be set in region to be detected in embodiments of the present invention, this region to be detected is monitored, the video information in this region to be detected can be obtained by photographic head.
To be the picture remained unchanged in a long time in video information be background in embodiments of the present invention, such as when carrying out the statistics of number in market, the decoration in market, fixing bunk are it is believed that background in this video information, and the object moved in this market, such as people, owing to it exists in this video information at short notice, it is believed that be sport foreground.
S102: each frame of video identified is carried out binary conversion treatment, after binary conversion treatment, in each frame of video, sport foreground is white pixel point, and background is black pixel point.
In order to facilitate follow-up number identification, being required for each frame of video in embodiments of the present invention and carry out binary conversion treatment, after binary conversion treatment, the personage in sport foreground can become white pixel point, and background can become black pixel point.
S103: for each frame of video after binary conversion treatment, detect each white pixel point in this frame of video, the pixel value according to the pixel adjacent with this white pixel point, it is determined that be positioned at the white pixel point on skeleton.
In embodiments of the present invention by each frame of video is carried out binary conversion treatment, after binaryzation, sport foreground becomes white pixel point, it is analyzed for each white pixel point, it is determined that whether each white pixel point is the white pixel point on skeleton, thus realizing " thinization " of people is processed.
Concrete when determining that whether each white pixel point is the white pixel point on skeleton, needing to refer to the pixel value of the pixel adjacent with this white pixel point, the pixel adjacent with this white pixel point includes pixel adjacent with this white pixel point in any direction.This white pixel point that Fig. 2 provides for the embodiment of the present invention and the position relationship schematic diagram of the pixel adjacent with this white pixel point, wherein p1For this white pixel point, p2～p9For the pixel adjacent with this white pixel point.
S104: according to the white pixel being positioned on the skeleton point determined, and the pixel value of adjacent each pixel, it is determined that whether this white pixel point is the pixel of the head region of people.
After determining the pixel being positioned on skeleton, for each pixel being positioned on skeleton, and the pixel value of each pixel being adjacent, it is judged which pixel is in the pixel of the head region of people.Due in embodiments of the present invention, in order to improve the efficiency of statistics, when the pixel of head region that the white pixel point determined on which skeleton is people, it is possible in accordance with the order from top to bottom, judge for each white pixel point successively.And, owing to the general difference of the size of the head of people is little, the quantity of the pixel that the head of people comprises can be set, from the pixel being arranged in skeleton, the pixel of number respective numbers downwards, said method is begun with, it is determined that whether this white pixel point is the pixel of the head region of people from next pixel.
S105: the pixel determining the head region for people is tracked, according to the result statistical number of person followed the tracks of.
After determining the pixel of head region of people, in order to improve the accuracy of demographics further, in embodiments of the present invention after each pixel that the head region determining people comprises, the pixel of the head region of people is tracked, determine whether for real people, and according to the final result followed the tracks of, carry out the statistics of number.
After determining the white pixel point being positioned on skeleton in embodiments of the present invention, it is achieved that " thinization " of human body is processed, it is possible to the interpersonal adhesion problems of effective solution, thus improve the accuracy of demographics.Additionally, in embodiments of the present invention when, after the pixel of head region determining people, the pixel of the head region of people being tracked, according to the number that the result statistics followed the tracks of is final, such that it is able to filter out error message, further increase the accuracy of demographics.
For the video information obtained, identify the background in this video information, namely relatively remain unchanged, or metastable information, or the information of relative quiescent, it is also possible to it is called environmental information.Owing to a video information is generally based on a kind of background, such as can video information in bank, or the video information in market, or other are public, video information in public place of entertainment, and owing to the background in video information is relatively fixed, therefore can first by the Background Recognition in video information out.Utilizing Gaussian Background modeling technique in embodiments of the present invention, by the background of single Gauss or many Gaussian Background modeling technique identification video information, the technology of this Background Recognition belongs to prior art, does not just repeat one by one with regard to this.
When after the background that have identified in video information, owing to the background in each frame of video is relatively fixing, according to the background identified, it is possible to extract the sport foreground in each frame of video.
Binary conversion treatment is carried out for each frame of video, owing to may determine that sport foreground and the background of this frame of video for each frame of video, therefore when carrying out binary conversion treatment, sport foreground in each frame of video is processed for white, background process is black, namely after binaryzation, the sport foreground in each frame of video is white pixel point, and background is black pixel point.
After each frame of video is carried out binary conversion treatment, in each frame of video, the pixel on skeleton can be identified, but in embodiments of the present invention in order to improve the recognition efficiency of skeleton pixel, and improve the accuracy of follow-up demographics, after each frame of video is carried out binary conversion treatment, denoising operation can also be carried out for each frame of video.
Concrete, Morphological scale-space can be carried out for each frame of video in embodiments of the present invention, namely each frame of video be carried out corrosion treatmentCorrosion Science, it can in addition contain carry out expansion process, thus reaching to remove Gaussian Background in frame of video to model the noise brought.
After removing the noise in each frame of video, being namely positioned at the white pixel point on skeleton in recognizable each frame of video, the concrete method determining the white pixel being positioned on skeleton point in embodiments of the present invention includes:
For each white pixel point, it is judged that in the pixel being adjacent, whether the quantity of white pixel point meets the first condition set；
In a certain order, it is judged that by whether the number of times that black pixel point saltus step is white pixel point meets the second condition set in each pixel being adjacent；
Judge whether the pixel value of the pixel being adjacent meets the third condition set；
When the conditions are satisfied, it is determined that the pixel on this white pixel non-human skeleton of point, otherwise, it determines this white pixel point is the white pixel point being positioned on skeleton.
In the embodiment of the present invention when determining the white pixel point being positioned on skeleton, it is required for each white pixel point to judge, judge whether this white pixel point is in the white pixel point on skeleton, and in the process judged, in addition it is also necessary to reference to other each pixels adjacent with this white pixel point.As in figure 2 it is shown, when this white pixel point is p1Time, it is necessary to reference and this p1Adjacent pixel p2～p9It is, concrete that need to refer in embodiments of the present invention is pixel p2～p9Pixel value.
Process to realize " thinization " to personage, have only to identify the position of skeleton in embodiments of the present invention, namely identify which the pixel constituting skeleton includes, because after in embodiments of the present invention frame of video being carried out binaryzation, sport foreground is white pixel point, is also white pixel point therefore in the pixel on skeleton.
When for each white pixel point, judge when whether this white pixel point is be positioned at the white pixel point on skeleton, it is required for multiple condition to judge, wherein, first condition is the quantity of white pixel point in the pixel adjacent with this white pixel point, when in the pixel adjacent with this white pixel point, the quantity of white pixel point meets the first condition of setting, it is also possible to think that this white pixel point meets this first condition.Specifically in embodiments of the present invention, it is possible to B (p1) represent and white pixel point p1The quantity of white pixel point in adjacent pixel, as this B (p1) when meeting between 2～6, it is believed that this white pixel point meets the first condition of setting, namely meets 2≤B (p1During)≤6, it is believed that this white pixel point p1Meet the first condition of this setting.
Need in a certain order in embodiments of the present invention, it is judged that by whether the number of times that black pixel point saltus step is white pixel point meets the second condition set in the pixel adjacent with this white pixel point, as shown in Figure 2, it is possible to according to p2～p9Order, judge pixel p successively2～p3、p3～p4、p4～p5、……、p9～p2In, by the number of times that black pixel point saltus step is white pixel point.Such as pixel p1For white pixel point, pixel p2～p9Respectively black pixel point, black pixel point, white pixel point, black pixel point, black pixel point, black pixel point, white pixel point, white pixel point, according to p2～p3、p3～p4、p4～p5、……、p9～p2Order, it is determined that be 2 by the transition times that black pixel point saltus step is white pixel point.
Adopt A (p in embodiments of the present invention1) represent in a certain order, with white pixel point p1By the transition times that black pixel point saltus step is white pixel point in adjacent each pixel.In embodiments of the present invention when in a certain order, in each pixel adjacent with white pixel point, during by the transition times that black pixel point saltus step is white pixel point equal to 1, the second condition thought and met setting in its adjacent each pixel of this white pixel point by the transition times that black pixel point saltus step is white pixel point, it is also possible to think this white pixel point p1Meet the second condition of this setting.
When determining whether each white pixel point is be positioned at the white pixel point on skeleton, in addition it is also necessary to determine whether the third condition meeting setting according to the pixel value of the pixel adjacent with each white pixel point.Need to judge whether the pixel of the upper and lower, left and right four direction of this white pixel point exists black pixel point in embodiments of the present invention.
As in figure 2 it is shown, namely judge and this white pixel point p1Adjacent pixel p2、p4、p6、p8In whether there is black pixel point, owing to the pixel value of black pixel point is 0, in judge process, it can be determined that pixel p2、p4、p6The product of pixel value whether be 0, and, it is judged that pixel p2、p4、p8The product of pixel value whether be 0, namely judge p2×p4×p6=0, and p4×p6×p8=0；Or, it is judged that pixel p2、p4、p8The product of pixel value whether be 0, and, it is judged that pixel p2、p6、p8The product of pixel value whether be 0, namely judge p2×p4×p8=0, and p2×p6×p8=0.
In embodiments of the present invention for white pixel point p1, it is judged that with this white pixel point at four, upper and lower, left and right pixel p2、p4、p6、p8Pixel value whether meet following condition:
p2×p4×p6=0, and p4×p6×p8=0；Or
p2×p4×p8=0, and p2×p6×p8=0.
When for each white pixel point, when the pixel adjacent with this white pixel point meets the first condition of above-mentioned setting, second condition and third condition, then determine this non-pixel being positioned on skeleton of white pixel point, otherwise, it determines this pixel is the white pixel point being positioned on skeleton.
After determining the pixel in order to be positioned on skeleton, can directly be tracked according to the pixel being positioned on skeleton, but in order to improve the accuracy of demographics in the embodiment of the present invention, realize by following the tracks of the head region of people, such that it is able to the interpersonal adhesion problems of significantly more efficient solution.
In embodiments of the present invention can at the white pixel point being arranged on skeleton, according to the white pixel point being positioned on skeleton, and the pixel value of adjacent each pixel, it is determined that whether each white pixel point is the pixel of the head region of people.The region at the head place according to people, in accordance with the order from top to bottom, can judge that whether each white pixel being positioned on skeleton point is the pixel of the head region of people in embodiments of the present invention successively.
Or, the quantity of the pixel that can also comprise according to the head of common people in embodiments of the present invention, white pixel point after corresponding quantity judges in accordance with the order from top to bottom, it is judged that whether this white pixel point is the pixel of the head region of people.Such as it is generally acknowledged that the head of people in the picture comprises the pixel of more than 10, then when judging, according to top to bottm in order, proceeding by judgement from the 11st pixel.
Concrete, determine that whether this white pixel point is that the pixel of the head region of people includes in embodiments of the present invention:
Add up the number of white pixel point in the pixel adjacent with this white pixel point, when the number of white pixel point meets the fourth condition of setting in the pixel that it is adjacent, it is determined that this white pixel point is the preliminary election pixel of the head region of people；
For each preliminary election pixel, it is judged that below this preliminary election pixel, in rectangular area, whether the quantity of white pixel point has exceeded setting threshold value；
When in this rectangular area, the quantity of white pixel point exceedes setting threshold value, it is determined that the pixel of the head region that this preliminary election pixel is behaved.
In embodiments of the present invention for it has been determined that each white pixel point being positioned on skeleton, it is judged that whether this white pixel point is the pixel of the head region of people.Concrete, in embodiments of the present invention for each white pixel being positioned on skeleton point, judge the quantity of white pixel point in the pixel adjacent with this white pixel point, when in the pixel adjacent with this white pixel point, the quantity of white pixel point is 1, think the candidate pixel point to the head region that white pixel point is people, otherwise, the candidate pixel point of the head region that this white pixel point is inhuman.
In each candidate pixel point determining the head region for people, for each candidate pixel point, construct rectangle in the underface of this candidate pixel point.As in figure 2 it is shown, work as p1During for candidate pixel point, with p6And p7The midpoint that center is long limit, rectangle upper end structure rectangle, or with p6And p5The midpoint that center is long limit, rectangle upper end structure rectangle.The concrete rectangle constructed in embodiments of the present invention is 20*10, namely comprises 20 pixels in the longitudinal direction, comprises 10 pixels in the direction of the width.
The each pixel comprised in this rectangle according to structure, add up the quantity of the white pixel point comprised in this rectangle, when the quantity of the white pixel point comprised in this rectangle has exceeded setting threshold value, then think the pixel of head region that this candidate pixel point is people, otherwise, the pixel of the head region that this candidate pixel point is inhuman.
In embodiments of the present invention when, after the pixel determining the head region being positioned at people, the pixel of the head region being pointed to people is tracked, according to the result statistical number of person followed the tracks of.Concrete is tracked the pixel determining the head region for people in embodiments of the present invention, and the result statistical number of person according to following the tracks of includes:
For the pixel determining the head region for people in each frame of video, and the head size scope arranged, it is determined that the region at the head place of people in each frame of video；
For the information of everyone head region in this frame of video, and the information of everyone head region in a upper frame of video, it is judged that there is overlap in the head region of the people whether the head region of this people is corresponding with a upper frame of video；
When existing overlapping, the first trajectory range that the information of the head region of people corresponding in the information updating of the head region of this people in this frame of video to a upper frame of video is preserved, when being absent from overlap, generate the second trajectory range, by the information updating of the head region of this people in this frame of video to this second trajectory range；
When the 3rd trajectory range in a upper frame of video does not update the information of head region of people in this frame of video, statistical number of person is added one, and delete the information preserved in the 3rd trajectory range.
In embodiments of the present invention, according to the pixel of the head region of people in each frame of video determined, the pixel of the head region of people is tracked, according to the result statistical number of person followed the tracks of.Concrete in embodiments of the present invention for after the pixel of the head region of the people determined in each frame of video, according to the head size scope arranged, determine the region at the head place of people in each frame of video, concrete in embodiments of the present invention for these pixels, constructing the length of side centered by each pixel is the square of 40, such that it is able to determine the region at everyone head place in each frame of video.
For each frame of video, according to everyone region at head place posttectonic, and the region at everyone head place in a upper frame of video, it is judged that the region at everyone head place in this frame of video, if there is overlap with the region at the head place of the corresponding people in a upper frame of video.Specifically in embodiments of the present invention, can according to the region at everyone the head place identified in first frame of video, owing to being absent from a frame of video before first frame of video, in this first frame of video, the region at everyone head place and a upper frame of video are absent from overlapping region, hence for the Area generation trajectory range at everyone the head place in first frame of video, the information of everyone head region in each first frame of video is saved in the trajectory range of correspondence.
For each frame of video afterwards, region for the head place of the people identified in each frame of video, and the region at everyone head place in a upper frame of video, it is judged that in this frame of video there is overlap in the region at the head place of the people whether region at the head place of this people is corresponding with a upper frame of video.Coordinate according to the head region of the people identified in frame of video in embodiments of the present invention, and the coordinate of the head region of people corresponding to a upper frame of video determines whether overlap.Additionally, in embodiments of the present invention when the region at the head place of people a certain in this frame of video, the overlapping area of the head region of the people corresponding with a upper frame of video is not less than the 30% of the head region of this people, then it is assumed that the head region of the people that the region at the head place of this people is corresponding with a upper frame of video is overlapping.
When the head region determining a certain people in this frame of video, with in a upper frame of video a certain people head region exist overlapping time, in the first trajectory range that the head region information of this correspondence people in the information updating of the head region of this people in this frame of video to a upper frame of video is preserved, it is saved in the first trajectory range by the information of the head region of this people in this frame of video frame；Head region as people a certain in this frame of video, with when in a upper frame of video, the head region of a certain people is absent from overlapping, then it is believed that corresponding artificial the incipient people of this frame of video in the head region of this people in this frame of video, it is required for the head place of this people when one new trajectory range of Area generation, i.e. the second trajectory range, by the information updating of the head region of this people in this frame of video to this second trajectory range.
Owing in a upper frame of video, the information of everyone head region is saved in the trajectory range of correspondence, after the region at everyone head place is mated with a upper frame of video in for this frame of video, when a upper frame of video exists the 3rd trajectory range not being updated the head region that storage has people, illustrate that somebody is not in this detection range in this frame of video, now statistical number of person is added one, and delete the information preserved in the 3rd trajectory range.
Owing to determining the white pixel point on skeleton in embodiments of the present invention, it is achieved that " thinization " of human body is processed, it is possible to the interpersonal adhesion problems of effective solution, thus improve the accuracy of demographics.
The structural representation of a kind of people counting device based on video that Fig. 3 provides for the embodiment of the present invention, described statistic device includes:
Identification module 31 treats, for obtaining, the video information that detection region carries out detecting, and identifies the sport foreground of each frame of video in this video information；The each frame of video identified is carried out binary conversion treatment, and after binary conversion treatment, in each frame of video, sport foreground is white pixel point, and background is black pixel point；
First determines module 32, for for each frame of video after binary conversion treatment, detecting each white pixel point in this frame of video, the pixel value according to the pixel adjacent with this white pixel point, it is determined that be positioned at the white pixel point on skeleton；
Second determines module 33, for according to the white pixel being positioned on the skeleton point determined, and the pixel value of adjacent each pixel, it is determined that whether this white pixel point is the pixel of the head region of people；
Tracking module 34, for being tracked the pixel determining the head region for people, according to the result statistical number of person followed the tracks of.
Described device also includes:
Pretreatment module 35, for carrying out corrosion treatmentCorrosion Science and expansion process to each frame of video.
Described first determines module 32, specifically for for each white pixel point, it is judged that in the pixel being adjacent, whether the quantity of white pixel point meets the first condition set；In a certain order, it is judged that in each pixel being adjacent, whether the transition times of pixel value meets the second condition set；Judge whether the pixel value of the pixel being adjacent meets the third condition set；When the conditions are satisfied, it is determined that the pixel on this white pixel non-human skeleton of point, otherwise, it determines this white pixel point is the white pixel point being positioned on skeleton.
Described second determines module 33, specifically for the number of white pixel point in the pixel that statistics is adjacent with this white pixel point, when in the pixel that it is adjacent, the number of white pixel point meets the fourth condition of setting, it is determined that this white pixel point is the preliminary election pixel of the head region of people；For each preliminary election pixel, it is judged that below this preliminary election pixel, in rectangular area, whether the quantity of white pixel point has exceeded setting threshold value；When in this rectangular area, the quantity of white pixel point exceedes setting threshold value, it is determined that the pixel of the head region that this preliminary election pixel is behaved.
Described tracking module 34 is specifically for for the pixel determining the head region for people in each frame of video, and the head size scope arranged, it is determined that the region at the head place of people in each frame of video；For the information of everyone head region in this frame of video, and the information of everyone head region in a upper frame of video, it is judged that there is overlap in the head region of the people whether the head region of this people is corresponding with a upper frame of video；When existing overlapping, the first trajectory range that the information of the head region of people corresponding in the information updating of the head region of this people in this frame of video to a upper frame of video is preserved, when being absent from overlap, generate the second trajectory range, by the information updating of the head region of this people in this frame of video to this second trajectory range；When the 3rd trajectory range in a upper frame of video does not update the information of head region of people in this frame of video, statistical number of person is added one, and delete the information preserved in the 3rd trajectory range.
The embodiment of the present invention provides a kind of demographic method based on video and device, the method includes: identify the sport foreground monitoring video information, and carry out binary conversion treatment for each frame of video, after binary conversion treatment, in each frame of video, sport foreground is white pixel point, background is black pixel point, for each white pixel point, pixel value according to other pixels adjacent with this white pixel point, determine the white pixel point being positioned on skeleton, and in these white pixel points, determine the pixel of the head region being positioned at people, pixel according to the head region being positioned at people determined is tracked, according to the result statistical number of person followed the tracks of.Owing to determining the white pixel point on skeleton in embodiments of the present invention, it is achieved that " thinization " of human body is processed, it is possible to the interpersonal adhesion problems of effective solution, thus improve the accuracy of demographics.
Obviously, the present invention can be carried out various change and modification without deviating from the spirit and scope of the present invention by those skilled in the art.So, if these amendments of the present invention and modification belong within the scope of the claims in the present invention and equivalent technologies thereof, then the present invention is also intended to comprise these change and modification.
Claims (6)
1. the demographic method based on video, it is characterised in that described statistical method includes:
Obtain and treat the video information that detection region carries out detecting, identify the sport foreground of each frame of video in this video information；
Each frame of video is carried out corrosion treatmentCorrosion Science and expansion process；
The each frame of video identified is carried out binary conversion treatment, and after binary conversion treatment, in each frame of video, sport foreground is white pixel point, and background is black pixel point；
For each frame of video after binary conversion treatment, detect each white pixel point in this frame of video, the pixel value according to the pixel adjacent with this white pixel point, it is determined that be positioned at the white pixel point on skeleton；
According to the white pixel being positioned on the skeleton point determined, and the pixel value of adjacent each pixel, it is determined that whether this white pixel point is the pixel of the head region of people；
The pixel determining the head region for people is tracked, according to the result statistical number of person followed the tracks of；
Wherein, described determine that whether this white pixel point is that the pixel of the head region of people includes:
Add up the number of white pixel point in the pixel adjacent with this white pixel point, when the number of white pixel point meets the fourth condition of setting in the pixel that it is adjacent, it is determined that this white pixel point is the preliminary election pixel of the head region of people；
For each preliminary election pixel, it is judged that below this preliminary election pixel, in rectangular area, whether the quantity of white pixel point has exceeded setting threshold value；
When in this rectangular area, the quantity of white pixel point exceedes setting threshold value, it is determined that the pixel of the head region that this preliminary election pixel is behaved.
2. statistical method as claimed in claim 1, it is characterised in that described determine that the white pixel being positioned on skeleton point includes:
For each white pixel point, it is judged that in the pixel being adjacent, whether the quantity of white pixel point meets the first condition set；
Judge in the pixel around described each white pixel point, neighbor pixel is by whether the number of times that black pixel point saltus step is white pixel point meets the second condition set；
Judge whether the pixel value of the pixel being adjacent meets the third condition set；
When the conditions are satisfied, it is determined that the pixel on this white pixel non-human skeleton of point, otherwise, it determines this white pixel point is the white pixel point being positioned on skeleton.
3. statistical method as claimed in claim 1, it is characterised in that the described pixel to determining the head region for people is tracked, and the result statistical number of person according to following the tracks of includes:
For the pixel determining the head region for people in each frame of video, and the head size scope arranged, it is determined that the region at the head place of people in each frame of video；
For the information of everyone head region in this frame of video, and the information of everyone head region in a upper frame of video, it is judged that there is overlap in the head region of the people whether the head region of this people is corresponding with a upper frame of video；
When existing overlapping, the first trajectory range that the information of the head region of people corresponding in the information updating of the head region of this people in this frame of video to a upper frame of video is preserved, when being absent from overlap, generate the second trajectory range, by the information updating of the head region of this people in this frame of video to this second trajectory range；
When the 3rd trajectory range in a upper frame of video does not update the information of head region of people in this frame of video, statistical number of person is added one, and delete the information preserved in the 3rd trajectory range.
4. the people counting device based on video, it is characterised in that described statistic device includes:
Identification module, treats, for obtaining, the video information that detection region carries out detecting, identifies the sport foreground of each frame of video in this video information；The each frame of video identified is carried out binary conversion treatment, and after binary conversion treatment, in each frame of video, sport foreground is white pixel point, and background is black pixel point；
First determines module, for for each frame of video after binary conversion treatment, detecting each white pixel point in this frame of video, the pixel value according to the pixel adjacent with this white pixel point, it is determined that be positioned at the white pixel point on skeleton；
Pretreatment module, for carrying out corrosion treatmentCorrosion Science and expansion process to each frame of video；
Second determines module, for according to the white pixel being positioned on the skeleton point determined, and the pixel value of adjacent each pixel, it is determined that whether this white pixel point is the pixel of the head region of people；
Tracking module, for being tracked the pixel determining the head region for people, according to the result statistical number of person followed the tracks of；
Wherein, described second determines module, specifically for the number of white pixel point in the pixel that statistics is adjacent with this white pixel point, when in the pixel that it is adjacent, the number of white pixel point meets the fourth condition of setting, it is determined that this white pixel point is the preliminary election pixel of the head region of people；For each preliminary election pixel, it is judged that below this preliminary election pixel, in rectangular area, whether the quantity of white pixel point has exceeded setting threshold value；When in this rectangular area, the quantity of white pixel point exceedes setting threshold value, it is determined that the pixel of the head region that this preliminary election pixel is behaved.
5. statistic device as claimed in claim 4, it is characterised in that described first determines module, specifically for for each white pixel point, it is judged that in the pixel being adjacent, whether the quantity of white pixel point meets the first condition set；Judge in the pixel around described each white pixel point, neighbor pixel is by whether the number of times that black pixel point saltus step is white pixel point meets the second condition set；Judge whether the pixel value of the pixel being adjacent meets the third condition set；When the conditions are satisfied, it is determined that the pixel on this white pixel non-human skeleton of point, otherwise, it determines this white pixel point is the white pixel point being positioned on skeleton.
6. statistic device as claimed in claim 4, it is characterized in that, described tracking module, specifically for for the pixel determining the head region for people in each frame of video, and the head size scope arranged, it is determined that the region at the head place of people in each frame of video；For the information of everyone head region in this frame of video, and the information of everyone head region in a upper frame of video, it is judged that there is overlap in the head region of the people whether the head region of this people is corresponding with a upper frame of video；When existing overlapping, the first trajectory range that the information of the head region of people corresponding in the information updating of the head region of this people in this frame of video to a upper frame of video is preserved, when being absent from overlap, generate the second trajectory range, by the information updating of the head region of this people in this frame of video to this second trajectory range；
When the 3rd trajectory range in a upper frame of video does not update the information of head region of people in this frame of video, statistical number of person is added one, and delete the information preserved in the 3rd trajectory range.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310005046.8A CN103077380B (en) | 2013-01-07 | 2013-01-07 | A kind of demographic method based on video and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310005046.8A CN103077380B (en) | 2013-01-07 | 2013-01-07 | A kind of demographic method based on video and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103077380A CN103077380A (en) | 2013-05-01 |
| CN103077380B true CN103077380B (en) | 2016-06-29 |
Family
ID=48153904
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310005046.8A Expired - Fee Related CN103077380B (en) | 2013-01-07 | 2013-01-07 | A kind of demographic method based on video and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103077380B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103530988B (en) * | 2013-10-11 | 2017-01-04 | 吉林大学 | A kind of lawn is trampleed on and the method for early warning of the extent of injury |
| KR101557376B1 (en) * | 2014-02-24 | 2015-10-05 | 에스케이 텔레콤주식회사 | Method for Counting People and Apparatus Therefor |
| CN107516124B (en) * | 2016-06-17 | 2020-06-16 | 中国科学院深圳先进技术研究院 | Real-time people flow number counting method and system |
| CN106485265B (en) * | 2016-09-22 | 2020-01-17 | 深圳大学 | Image identification method and device |
| WO2018053763A1 (en) * | 2016-09-22 | 2018-03-29 | 深圳大学 | Image identification method and device |
| CN108830145B (en) * | 2018-05-04 | 2021-08-24 | 深圳技术大学（筹） | People counting method based on deep neural network and storage medium |
| CN113091227B (en) * | 2020-01-08 | 2022-11-01 | 佛山市云米电器科技有限公司 | Air conditioner control method, cloud server, air conditioner control system and storage medium |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102509070A (en) * | 2011-10-12 | 2012-06-20 | 西安理工大学 | Video-based human face area tracking method for counting people paying close attention to advertisement |
-
2013
- 2013-01-07 CN CN201310005046.8A patent/CN103077380B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102509070A (en) * | 2011-10-12 | 2012-06-20 | 西安理工大学 | Video-based human face area tracking method for counting people paying close attention to advertisement |
Non-Patent Citations (2)
| Title |
|---|
| 《基于视频图像处理的人数自动统计技术研究》;顾德军;《万方学位论文全文库》;20121211;第4.2.1-4.2.3节，第5.4.2.1-5.4.2.2节，第5.5节 * |
| 《基于骨架化和模板匹配的交通指挥手势识别》;李文杰;《万方学位论文全文库》;20111031;第2.3节，第4.4节，图2-2，图4-4 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103077380A (en) | 2013-05-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103077380B (en) | A kind of demographic method based on video and device | |
| CN108509859B (en) | Non-overlapping area pedestrian tracking method based on deep neural network | |
| CN106845487B (en) | End-to-end license plate identification method | |
| Yamaguchi et al. | Fast crack detection method for large-size concrete surface images using percolation-based image processing | |
| CN104217427B (en) | Lane line localization method in a kind of Traffic Surveillance Video | |
| CN110414559B (en) | Construction method of intelligent retail cabinet commodity target detection unified framework and commodity identification method | |
| Kalsotra et al. | Background subtraction for moving object detection: explorations of recent developments and challenges | |
| CN101470809B (en) | Moving object detection method based on expansion mixed gauss model | |
| CN108268867B (en) | License plate positioning method and device | |
| CN104794435B (en) | A kind of unmanned plane of view-based access control model moving target detecting method over the ground | |
| Lee et al. | A genetic algorithm-based moving object detection for real-time traffic surveillance | |
| US10157327B2 (en) | Image processing device, image processing method, and program | |
| CN106295532B (en) | A kind of human motion recognition method in video image | |
| KR101436369B1 (en) | Apparatus and method for detecting multiple object using adaptive block partitioning | |
| CN104599511B (en) | Traffic flow detection method based on background modeling | |
| CN104036243A (en) | Behavior recognition method based on light stream information | |
| CN110781844A (en) | Security patrol monitoring method and device | |
| CN103870824A (en) | Method and device for capturing face in face detecting and tracking process | |
| CN103955682A (en) | Behavior recognition method and device based on SURF interest points | |
| CN107590431B (en) | Quantity counting method and device based on image recognition | |
| CN113792606A (en) | Low-cost self-supervision pedestrian re-identification model construction method based on multi-target tracking | |
| CN112200020A (en) | Pedestrian re-identification method and device, electronic equipment and readable storage medium | |
| CN108648210B (en) | Rapid multi-target detection method and device under static complex scene | |
| CN103049749A (en) | Method for re-recognizing human body under grid shielding | |
| CN114998317A (en) | Lens occlusion detection method and device, camera device and storage medium |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20180615 Address after: 100096 Haidian District, Beijing, west 2 new apartment building, three floor commercial room 337. Patentee after: Letter frame technology (Beijing) Co., Ltd. Address before: 100085 A-6, 28, Beijing, Haidian District, Shanghai. Patentee before: Xinzheng Electronic Technology (Beijing) Co., Ltd. |
|
| TR01 | Transfer of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160629 Termination date: 20200107 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |