CN105303191A - Method and apparatus for counting pedestrians in foresight monitoring scene - Google Patents
Method and apparatus for counting pedestrians in foresight monitoring scene Download PDFInfo
- Publication number
- CN105303191A CN105303191A CN201410360781.5A CN201410360781A CN105303191A CN 105303191 A CN105303191 A CN 105303191A CN 201410360781 A CN201410360781 A CN 201410360781A CN 105303191 A CN105303191 A CN 105303191A
- Authority
- CN
- China
- Prior art keywords
- head
- head zone
- zone
- pedestrian
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C9/00—Individual registration on entry or exit
Abstract
The invention discloses a method for counting pedestrians in a foresight monitoring scene. The method comprises: setting the positions of a detection region and a detection line, performing head region and head-shoulder region detection on a motion mask image in the detection region, and determining a definite head region according to a position relationship between the head region and the head-shoulder region; and tracking a motion track of the determined head region, and according to the motion track and the position of the detection line, counting the pedestrians by direction. The invention furthermore discloses an apparatus for counting the pedestrians in the foresight monitoring scene.
Description
Technical field
The present invention relates to Video Supervision Technique, particularly relate to the pedestrian counting method under a kind of forward sight supervision scene and device.
Background technology
Along with the development of Video Supervision Technique, the people counting of view-based access control model is more and more subject to people's attention, and is applied to multiple occasion.People counting technology can play not same-action in different occasion, can the flow of guarder on scenic spot, park and other places, when the super warning of the flow of people, controls the flow of people, prevents because the crowded potential safety hazard caused; On market and other places, the time and space distribution of passenger flow can be determined, increase targetedly or reduce a staff, improving service level; At important places such as power plant, can access control, monitoring is abnormal, and alarm.
For people counting algorithm aspect, external scientific research institution is as CMU, Tl company, and the research tended to crowd density, wherein ripe effective method has dynamic texture model, perspective method and the method based on mathematical morphology more.But dynamic texture method computation complexity is higher, perspective method needs by mode identification method identification pedestrian, thus makes these two kinds of methods all be difficult to be applied to real-time video monitoring.Method based on mathematical morphology can meet real-time condition, but application scenarios overlooks video camera based on forward, causes extendability poor.In order to realize counting pedestrian's quantity by a certain fixed position based on time series, the method identification pedestrian that the people such as O.Sidla and Y.Lypetskyy adopt Shape-based interpolation identification to combine with target following and then statistical number of person, processing speed reaches 15 frames per second.For the dense population of railway station, subway station, the people such as A.N.Marana adopt the method extracting feature to solve density Estimation problem.Based on the thought of background subtraction, the people such as T.W.S.Chow adopt crowd's foreground area and edge feature to carry out the estimation of crowd density grade.
Number of patent application is the Chinese patent of 200910076256.X (notification number is CN101477641), describe a kind of demographic method based on video monitoring and system, this invention adopts Haar feature to carry out number of people detection, to count the tracking of the number of people and estimation.The deficiency that this invention exists is that the accuracy rate that the number of people detects is not high, and false drop rate is higher.Number of patent application is the Chinese patent of 201010114819.2 (notification number is CN101877058A), the method and system of a kind of people flow rate statistical introduced needs to train sorters different in a large number, such as light hair, dark hair, cap etc., again nose curve matching is carried out to these regions, training comparatively bothers, and can produce a large amount of flase drop.
Summary of the invention
For solving the technical matters of existing existence, the present invention mainly provides a kind of forward sight to monitor pedestrian counting method under scene and device.
Technical scheme of the present invention is achieved in that
The invention provides the pedestrian counting method under a kind of forward sight supervision scene, the method comprises:
Watch-dog arranges the position of surveyed area and detection line, and carry out head zone to the motion mask image in surveyed area, head shoulder regions detects, the position relationship according to head zone and head shoulder regions determines definite head zone; Follow the tracks of the movement locus of determined head zone, and according to described movement locus and detection line position, a point direction counts to pedestrian.
In such scheme, the described position arranging surveyed area and detection line comprises: on monitoring scene, arrange the surveyed area for demarcating sensing range, and in surveyed area, arranges the position of two detection lines according to the discrepancy direction of pedestrian.
In such scheme, described head zone is carried out to the motion mask image in surveyed area, head shoulder regions detects, position relationship according to head zone and head shoulder regions determines definite head zone, comprise: Gaussian modeling is carried out to supervision scene, extract the foreground area of motion, motion mask image is set up according to described foreground area, by head sorter and head-and-shoulder area class device, the head zone of same pedestrian is carried out respectively to the motion mask image in surveyed area, head shoulder regions detects, according to the head zone of same pedestrian, the geometric position restriction relation of head both shoulder regions determines the definite head zone of described pedestrian.
In such scheme, the movement locus of the determined head zone of described tracking comprises: pre-set tracking sequence, judge whether the head zone determined exists in tracking sequence, if existed, then upgrades current location and the trace template of described head zone; If there is no, then described head zone is added tracking sequence, and record the current location of described head zone, create the trace template of described head zone, described trace template is for recording the movement locus of head zone.
In such scheme, the method also comprises: obtain head-and-shoulder area class device and head sorter by the sample image of cascade classifier training head shoulder and head.
The invention provides the people counting device under a kind of forward sight supervision scene, this device comprises: detect and arrange module, number of people detection module, number of people tracking module, number of people counting module; Wherein,
Detection arranges module, for arranging the position of surveyed area and detection line;
Number of people detection module, for carrying out head zone to the motion mask image in surveyed area, head shoulder regions detects, and the position relationship according to head zone and head shoulder regions determines definite head zone;
Number of people tracking module, for following the tracks of the movement locus of described head zone;
Number of people counting module, for according to described movement locus and detection line position, a point direction counts pedestrian.
In such scheme, this device also comprises: motion masked areas extraction module, for carrying out Gaussian modeling to supervision scene, extracting the foreground area of motion, setting up motion mask image according to described foreground area.
In such scheme, this device also comprises: sorter training module, for obtaining head-and-shoulder area class device and head sorter by the sample image of cascade classifier training head shoulder and head.
In such scheme, described number of people detection module, specifically for carrying out the head zone of same pedestrian by head sorter and head-and-shoulder area class device respectively to the motion mask image in surveyed area, head shoulder regions detects, and determines the definite head zone of described pedestrian according to the head zone of same pedestrian, the geometric position restriction relation of both head shoulder regions.
In such scheme, described number of people tracking module, specifically for judging whether the head zone determined exists in tracking sequence, if existed, then upgrades current location and the trace template of described head zone; If there is no, then described head zone is added tracking sequence, and record the current location of described head zone, create the trace template of described head zone, described trace template is for recording the movement locus of head zone.
The invention provides the pedestrian counting method under a kind of forward sight supervision scene and device, the position of surveyed area and detection line is set, carry out head zone to the motion mask image in surveyed area, head shoulder regions detects, the position relationship according to head zone and head shoulder regions determines definite head zone; Follow the tracks of the movement locus of determined head zone, and according to described movement locus and detection line position, a point direction counts to pedestrian; So, accurately can record direction of motion and the track of each pedestrian at surveyed area, record the pedestrian's number in monitor area comparatively accurately, occlusion issue when well avoiding pedestrian more, and reduce computational complexity.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet that the present invention realizes that forward sight monitors the pedestrian counting method under scene;
Fig. 2 is the schematic flow sheet that the present invention realizes training head sorter and head-and-shoulder area class device;
Fig. 3 is Adaboost cascade classifier training process schematic diagram provided by the invention;
Fig. 4 is the extraction schematic diagram of motion mask image provided by the invention;
Fig. 5 is the schematic flow sheet that the present invention realizes head zone detection;
Fig. 6 is the schematic flow sheet that the present invention realizes head zone tracking;
Fig. 7 is the effect schematic diagram that the present invention realizes people counting;
Fig. 8 is the structural representation that the present invention realizes that forward sight monitors the people counting device under scene.
Embodiment
In the embodiment of the present invention, arrange the position of surveyed area and detection line, carry out head zone to the motion mask image in surveyed area, head shoulder regions detects, the position relationship according to head zone and head shoulder regions determines definite head zone; Follow the tracks of the movement locus of determined head zone, and according to described movement locus and detection line position, a point direction counts to pedestrian.
Below by drawings and the specific embodiments, the present invention is described in further detail.
The embodiment of the present invention realizes the pedestrian counting method under a kind of forward sight supervision scene, and as shown in Figure 1, the method comprises following step:
Step 101: watch-dog arranges the position of surveyed area and detection line;
Because monitoring scene scope is larger, watch-dog arranges the surveyed area for demarcating sensing range on monitoring scene, surveyed area can be reduced, and the position of two detection lines is set in surveyed area according to the discrepancy direction of pedestrian, the direction of two detection lines is passed according to the movement locus of pedestrian head, the direction of pedestrian can be determined, effectively can reduce the calculated amount of detection by arranging surveyed area and detection line, improving detection perform.
Step 102: watch-dog carries out head zone to the motion mask image in surveyed area, head shoulder regions detects, and the position relationship according to head zone and head shoulder regions determines definite head zone;
Concrete, watch-dog carries out Gaussian modeling to supervision scene, extract the foreground area of motion, motion mask image is set up according to described foreground area, by head sorter and head-and-shoulder area class device, the head zone of same pedestrian is carried out respectively to the motion mask image in surveyed area, head shoulder regions detects, determine the definite head zone of described pedestrian according to the head zone of same pedestrian, the geometric position restriction relation of both head shoulder regions.
Step 103: the movement locus of determined head zone followed the tracks of by watch-dog;
Concrete, watch-dog pre-sets tracking sequence, and described tracking sequence is for storing the head zone of different pedestrian, and whether the head zone determined in determining step 102 has existed in tracking sequence, if existed, then upgrade current location and the trace template of described head zone; If there is no, then described head zone is added tracking sequence, and record the current location of described head zone, create the trace template of described head zone, described trace template is for recording the movement locus of head zone.
Step 104: watch-dog is according to described movement locus and detection line position, and a point direction counts pedestrian.
In above-mentioned steps 102, described head sorter and head-and-shoulder area class device need training in advance to obtain, concrete steps as shown in Figure 2:
Step 201: prepare positive sample and negative sample;
Concrete, collect a large amount of head images and head shoulder image as positive sample, positive sample and negative sample, as negative sample, are normalized to unified size by non-head shoulder images, non-head image and background image, and align sample and carry out mask process.
Step 202: the HOG feature extracting positive sample and negative sample, is normalized the HOG feature extracted;
Step 203: carry out cascade classifier training to the HOG feature of the positive sample after normalized and negative sample, obtains head sorter and head-and-shoulder area class device;
Concrete, the HOG feature of the positive negative sample after step 202 gained normalized is carried out the training of Adaboost cascade classifier respectively, obtains head-and-shoulder area class device and head sorter respectively, for detecting head zone and head shoulder regions respectively; Described Adaboost cascade classifier training process as shown in Figure 3, wherein, sorter h1, sorter h2, sorter h3 cascade, there is n negative sample, in negative example base, randomly draw m the negative sample with positive sample equivalent, carry out the training of every first-level class device, increased weights by the negative sample that mistake is divided after every first-level class device training and join in the training of next stage sorter, until reach training progression, obtain head-and-shoulder area class device and head sorter.Here, require that negative sample quantity is abundant, therefore will prepare abundant negative sample image, sliding window shape formula can be adopted to produce negative sample at random.
In above-mentioned steps 102, described watch-dog carries out Gaussian modeling to supervision scene, and extract the foreground area of motion, setting up motion mask image according to described foreground area can be specifically:
Motion masked areas is extracted and is divided into background modeling and foreground extraction two parts, and background modeling adopts mixed Gauss model to carry out modeling, and foreground extraction adopts background subtraction method to extract sport foreground.
The mathematical description of mixed Gauss model is as follows:
Any pixel x, can be expressed as in the value of t
for
time ordered sets X
t={ x
(t)..., x
(t-T), due to X
tmiddlely comprise background pixel and foreground pixel, therefore by X simultaneously
tthe Density Estimator function drawn can be expressed as formula (1).
Wherein
by the X in a period of time
tthe Gaussian mixtures obtained is estimated, is made up of M single Gauss model,
represent that the Estimation of Mean of each single Gaussian distribution in Gaussian mixtures, variance evaluation and weight are estimated respectively,
for gaussian kernel.
Mixed Gauss model adopts on-line study, automatic undated parameter, introduces mixture model adaptation mechanism, dynamically updates Gauss model number.When pixel belongs to one of B background model, just described pixel is designated as background, otherwise is designated as prospect, thus obtain the bianry image of entire image:
The foreground area of motion is extracted according to bianry image, as shown in Figure 4, foreground image pixel value in moving region is retained, and background dot is set to unified numerical value, form motion mask image, it is to be noted that the value of background pixel point can not be similar to foreground area edge pixel point, foreground edge feature can be made like this to die down, be unfavorable for further detection and tracking.
In above-mentioned steps 102, describedly by head sorter and head-and-shoulder area class device, the head zone of same pedestrian is carried out respectively to the motion mask image in surveyed area, head shoulder regions detects, determine the head zone of described pedestrian according to the head zone of same pedestrian, the geometric position restriction relation of both head shoulder regions, can be specifically:
On the basis of motion mask image, set up image pyramid, on each tomographic image pyramidal, adopt the window of fixed size, slide on image with the step-length of setting, the head sorter utilizing off-line training good, head-and-shoulder area class device judge whether the region in sliding window is head or head shoulder respectively, and testing result are merged, and determine the head zone of pedestrian.Key step is as shown in Figure 5:
Step 501: image pyramid is set up to motion mask image;
In this step, because in the image of monitor video, pedestrian head size is not fixed, in order to adapt to multiple dimensioned head detection, multiple dimensioned image pyramid is set up to image, namely to set zoom factor, repeatedly convergent-divergent is carried out to image, form the multi-layer image of different size.
Step 502: pixel grey scale space compression is carried out to every tomographic image of image pyramid;
Here, the every tomographic image of method to image pyramid adopting Gamma to correct carries out pixel grey scale space compression, reduces shade and the illumination variation of described image local, and described Gamma compresses formula and is:
I(x，y)＝I(x，y)
gamma(3)
Wherein, I (x, y) represents the pixel value of image in coordinate (x, y) position, gets gamma=1/2 here.
Step 503: the image gradient calculating the every tomographic image after pixel grey scale space compression;
Gradient calculation is carried out to the image after the Gamma obtained in step 502 corrects, when to coloured image compute gradient, gradient and the direction of pixel to be calculated each passage respectively, choose the gradient of pixel in the maximum passage of wherein Grad and the direction result as described pixel.
Step 504: the image gradient according to every tomographic image forms integration histogram;
Concrete, the gradient direction of each pixel of the every tomographic image obtained in step 503 is divided into 9 directions, and namely each pixel forms a gradient orientation histogram, then adopts the mode of integrogram to form integration histogram.
Step 505: HOG feature extraction is carried out to the integration histogram of every tomographic image;
Concrete, to the integration histogram of the every tomographic image obtained in step 504, adopt the window being applicable to head shoulder, head sizes respectively, with the step-length set in the enterprising line slip of integration histogram, calculate the HOG feature in each sliding window.
Step 506: utilize head-and-shoulder area class device, head shoulder regions that head sorter sorts out every tomographic image and head zone;
Concrete, head-and-shoulder area class device is adopted respectively to the head shoulder HOG feature obtained in step 505, head H OG feature, head sorter classifies, obtain head shoulder regions and the head zone of every tomographic image.
Step 507: the head shoulder regions of every tomographic image and head zone are merged, obtains the definite head zone of this tomographic image;
Because the head zone of same people must be positioned at its shoulder regions, therefore, whether the head shoulder regions that can obtain in determining step 506, head zone exist relation of inclusion, the head shoulder regions and head zone that there is relation of inclusion are merged and obtains final definite head zone, filter the head shoulder regions and head zone that there is not relation of inclusion.
Step 508: the definite head zone of each tomographic image is merged, obtains the definite head zone of motion mask image;
Concrete, carry out multiple dimensioned merging according to the size of the head zone of every tomographic image and position, obtain the definite head zone of final motion mask image.
In order to reach real-time and obtain good tracking effect, take here to detect once every 5 frames, the mode that residue frame is followed the tracks of, judge whether present frame detects according to frame number, the frame carrying out detecting is called detection frame, and the frame carrying out following the tracks of is called tracking frame.Above-mentioned steps 103 is concrete as shown in Figure 6:
Step 601: the frame number and the image information that obtain present frame;
Step 602: judging that present frame is as detecting frame or tracking frame, when for detecting frame, performs step 603 according to frame number, when for tracking frame, performing step 607;
Step 603: obtain the head zone detected;
Step 604: judge whether the head zone detected exists in tracking sequence, if existed, then performs step 605, if there is no, then performs step 606;
Due to head size between two frames and change in location less, therefore can by detecting that the size and location of head zone with the head zone existed in tracking sequence compare, if be less than decision threshold, for exist in tracking sequence, otherwise do not exist in tracking sequence.
Step 605: the current location and the trace template that upgrade described head zone, process ends;
Step 606: described head zone is added tracking sequence, records the current location of described head zone, creates the trace template of described head zone, and described trace template is for recording the movement locus of head zone, process ends;
Step 607: open search window near previous frame head zone, adopts three step search algorithm to mate trace model, obtains the matching value of best match position and best match position;
Step 608: described matching value and threshold value are compared;
Described threshold value is what pre-set, can be the number percent of matching degree.
Step 609: if matching value is greater than threshold value, process ends;
Step 610: if matching value is less than threshold value, then as final tracking results;
Step 611: the position determining the head zone of present frame, upgrades current location and the trace template of described head zone.
In step 607, if head zone do not detected when mating trace model, then by the current location of the tracing positional prediction head zone of previous frame.
If the head zone in tracking sequence is not again detected for a long time in surveyed area, be then considered to decoy, it is deleted in tracking sequence.
For step 104 can be:
During people counting, first at counting first frame, counter is reset, then the movement locus of head zone according to pedestrian, judge pedestrian movement direction, as shown in Figure 7, in surveyed area, count with the pedestrian of different motion direction by detection line.
In order to realize said method, the people counting device under the embodiment of the present invention provides a kind of forward sight to monitor scene, as shown in Figure 8, this device comprises: detect and arrange module 801, number of people detection module 802, number of people tracking module 803, number of people counting module 804; Wherein,
Detection arranges module 801, for arranging the position of surveyed area and detection line;
Number of people detection module 802, for carrying out head zone to the motion mask image in surveyed area, head shoulder regions detects, and the position relationship according to head zone and head shoulder regions determines definite head zone;
Number of people tracking module 803, for following the tracks of the movement locus of determined head zone;
Number of people counting module 804, for according to described movement locus and detection line position, a point direction counts pedestrian.
This device also comprises: motion masked areas extraction module 805, for carrying out Gaussian modeling to supervision scene, extracting the foreground area of motion, setting up motion mask image according to described foreground area;
This device also comprises: sorter training module 806, for obtaining head-and-shoulder area class device and head sorter by the sample image of cascade classifier training head shoulder and head;
Described number of people detection module 802, specifically for carrying out the head zone of same pedestrian by head sorter and head-and-shoulder area class device respectively to the motion mask image in surveyed area, head shoulder regions detects, and determines the definite head zone of described pedestrian according to the head zone of same pedestrian, the geometric position restriction relation of both head shoulder regions;
Described number of people tracking module 803, specifically for judging whether determined head zone exists in tracking sequence, if existed, then upgrades current location and the trace template of described head zone; If there is no, then described head zone is added tracking sequence, and record the current location of described head zone, create the trace template of described head zone, described trace template is for recording the movement locus of head zone.
The each embodiment of comprehensive the invention described above, the definite head zone of each pedestrian can be determined further by the position relationship of the head zone that detects and head shoulder regions at surveyed area, accurately record direction of motion and the track of each pedestrian, have recorded the pedestrian's number in monitor area more accurately, occlusion issue when well avoiding pedestrian more, and reduce computational complexity.
The above, be only preferred embodiment of the present invention, be not intended to limit protection scope of the present invention, and all any amendments done within the spirit and principles in the present invention, equivalent replacement and improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. the pedestrian counting method under forward sight supervision scene, it is characterized in that, the method comprises:
Watch-dog arranges the position of surveyed area and detection line, and carry out head zone to the motion mask image in surveyed area, head shoulder regions detects, the position relationship according to head zone and head shoulder regions determines definite head zone; Follow the tracks of the movement locus of determined head zone, and according to described movement locus and detection line position, a point direction counts to pedestrian.
2. pedestrian counting method according to claim 1, it is characterized in that, the described position arranging surveyed area and detection line comprises: on monitoring scene, arrange the surveyed area for demarcating sensing range, and in surveyed area, arranges the position of two detection lines according to the discrepancy direction of pedestrian.
3. pedestrian counting method according to claim 1, it is characterized in that, described head zone is carried out to the motion mask image in surveyed area, head shoulder regions detects, position relationship according to head zone and head shoulder regions determines definite head zone, comprise: Gaussian modeling is carried out to supervision scene, extract the foreground area of motion, motion mask image is set up according to described foreground area, by head sorter and head-and-shoulder area class device, the head zone of same pedestrian is carried out respectively to the motion mask image in surveyed area, head shoulder regions detects, according to the head zone of same pedestrian, the geometric position restriction relation of head both shoulder regions determines the definite head zone of described pedestrian.
4. pedestrian counting method according to claim 3, it is characterized in that, the movement locus of the determined head zone of described tracking comprises: pre-set tracking sequence, judge whether the head zone determined has existed in tracking sequence, if existed, then upgrade current location and the trace template of described head zone; If there is no, then described head zone is added tracking sequence, and record the current location of described head zone, create the trace template of described head zone, described trace template is for recording the movement locus of head zone.
5. pedestrian counting method according to claim 3, is characterized in that, the method also comprises: obtain head-and-shoulder area class device and head sorter by the sample image of cascade classifier training head shoulder and head.
6. the people counting device under forward sight supervision scene, it is characterized in that, this device comprises: detect and arrange module, number of people detection module, number of people tracking module, number of people counting module; Wherein,
Detection arranges module, for arranging the position of surveyed area and detection line;
Number of people detection module, for carrying out head zone to the motion mask image in surveyed area, head shoulder regions detects, and the position relationship according to head zone and head shoulder regions determines definite head zone;
Number of people tracking module, for following the tracks of the movement locus of described head zone;
Number of people counting module, for according to described movement locus and detection line position, a point direction counts pedestrian.
7. people counting device according to claim 6, it is characterized in that, this device also comprises: motion masked areas extraction module, for carrying out Gaussian modeling to supervision scene, extract the foreground area of motion, set up motion mask image according to described foreground area.
8. people counting device according to claim 7, is characterized in that, this device also comprises: sorter training module, for obtaining head-and-shoulder area class device and head sorter by the sample image of cascade classifier training head shoulder and head.
9. people counting device according to claim 8, it is characterized in that, described number of people detection module, specifically for carrying out the head zone of same pedestrian by head sorter and head-and-shoulder area class device respectively to the motion mask image in surveyed area, head shoulder regions detects, and determines the definite head zone of described pedestrian according to the head zone of same pedestrian, the geometric position restriction relation of both head shoulder regions.
10. people counting device according to claim 9, it is characterized in that, described number of people tracking module, specifically for judging whether the head zone determined has existed in tracking sequence, if existed, then upgrade current location and the trace template of described head zone; If there is no, then described head zone is added tracking sequence, and record the current location of described head zone, create the trace template of described head zone, described trace template is for recording the movement locus of head zone.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410360781.5A CN105303191A (en) | 2014-07-25 | 2014-07-25 | Method and apparatus for counting pedestrians in foresight monitoring scene |
| PCT/CN2015/072048 WO2015131734A1 (en) | 2014-07-25 | 2015-01-30 | Method, device, and storage medium for pedestrian counting in forward looking surveillance scenario |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410360781.5A CN105303191A (en) | 2014-07-25 | 2014-07-25 | Method and apparatus for counting pedestrians in foresight monitoring scene |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105303191A true CN105303191A (en) | 2016-02-03 |
Family
ID=54054576
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410360781.5A Withdrawn CN105303191A (en) | 2014-07-25 | 2014-07-25 | Method and apparatus for counting pedestrians in foresight monitoring scene |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN105303191A (en) |
| WO (1) | WO2015131734A1 (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326851A (en) * | 2016-08-19 | 2017-01-11 | 杭州智诺科技股份有限公司 | Head detection method |
| CN107092915A (en) * | 2016-02-18 | 2017-08-25 | 中国移动通信集团浙江有限公司 | A kind of method and apparatus for detecting crowd density |
| CN107403137A (en) * | 2017-06-29 | 2017-11-28 | 山东师范大学 | Dense population method of calculating flux and device based on video |
| CN108197579A (en) * | 2018-01-09 | 2018-06-22 | 杭州智诺科技股份有限公司 | The detection method of number in protective cabin |
| CN108280427A (en) * | 2018-01-24 | 2018-07-13 | 成都鼎智汇科技有限公司 | A kind of big data processing method based on flow of the people |
| CN108345842A (en) * | 2018-01-24 | 2018-07-31 | 成都鼎智汇科技有限公司 | A kind of processing method based on big data |
| CN110490030A (en) * | 2018-05-15 | 2019-11-22 | 保定市天河电子技术有限公司 | A kind of channel demographic method and system based on radar |
| CN110705408A (en) * | 2019-09-23 | 2020-01-17 | 东南大学 | Indoor people counting method and system based on mixed Gaussian people distribution learning |
| CN111091529A (en) * | 2018-10-24 | 2020-05-01 | 株式会社理光 | People counting method and system |
| CN111797652A (en) * | 2019-04-09 | 2020-10-20 | 佳能株式会社 | Object tracking method, device and storage medium |
| CN112084959A (en) * | 2020-09-11 | 2020-12-15 | 腾讯科技（深圳）有限公司 | Crowd image processing method and device |
| CN112232210A (en) * | 2020-10-16 | 2021-01-15 | 京东方科技集团股份有限公司 | Personnel flow analysis method and system, electronic device and readable storage medium |
| CN112446340A (en) * | 2020-12-07 | 2021-03-05 | 深圳市信义科技有限公司 | Pedestrian searching method and system combining pedestrian local feature and decorative attribute classification and storage medium |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108509896B (en) * | 2018-03-28 | 2020-10-13 | 腾讯科技（深圳）有限公司 | Trajectory tracking method and device and storage medium |
| CN110210302B (en) * | 2019-04-26 | 2023-06-20 | 平安科技（深圳）有限公司 | Multi-target tracking method, device, computer equipment and storage medium |
| CN110245556A (en) * | 2019-05-06 | 2019-09-17 | 深圳耄耋看护科技有限公司 | A kind of detection method, device, system and the storage medium of the state of leaving home of being in |
| CN110276778B (en) * | 2019-05-08 | 2022-10-28 | 西藏民族大学 | Animal circle-entering track extraction and statistical model construction and statistical method and device |
| CN110889339B (en) * | 2019-11-12 | 2020-10-02 | 南京甄视智能科技有限公司 | Head and shoulder detection-based dangerous area grading early warning method and system |
| CN111160203B (en) * | 2019-12-23 | 2023-05-16 | 中电科新型智慧城市研究院有限公司 | Loitering and stay behavior analysis method based on head-shoulder model and IOU tracking |
| CN111723664A (en) * | 2020-05-19 | 2020-09-29 | 烟台市广智微芯智能科技有限责任公司 | Pedestrian counting method and system for open type area |
| CN111652900B (en) * | 2020-05-29 | 2023-09-29 | 浙江大华技术股份有限公司 | Method, system and equipment for counting passenger flow based on scene flow and storage medium |
| CN112052838A (en) * | 2020-10-10 | 2020-12-08 | 腾讯科技（深圳）有限公司 | Object flow data monitoring method and device and readable storage medium |
| CN113807185B (en) * | 2021-08-18 | 2024-02-27 | 苏州涟漪信息科技有限公司 | Data processing method and device |
| CN113674303A (en) * | 2021-08-31 | 2021-11-19 | Oppo广东移动通信有限公司 | Image processing method, image processing device, electronic equipment and storage medium |
| CN117132942B (en) * | 2023-10-20 | 2024-01-26 | 山东科技大学 | Indoor personnel real-time distribution monitoring method based on region segmentation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101577812A (en) * | 2009-03-06 | 2009-11-11 | 北京中星微电子有限公司 | Method and system for post monitoring |
| 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 |
| CN101872422A (en) * | 2010-02-10 | 2010-10-27 | 杭州海康威视软件有限公司 | People flow rate statistical method and system capable of precisely identifying targets |
| CN102568005A (en) * | 2011-12-28 | 2012-07-11 | 江苏大学 | Moving object detection method based on Gaussian mixture model |
| US20120274781A1 (en) * | 2011-04-29 | 2012-11-01 | Siemens Corporation | Marginal space learning for multi-person tracking over mega pixel imagery |
| CN102799935A (en) * | 2012-06-21 | 2012-11-28 | 武汉烽火众智数字技术有限责任公司 | Human flow counting method based on video analysis technology |
| CN103021059A (en) * | 2012-12-12 | 2013-04-03 | 天津大学 | Video-monitoring-based public transport passenger flow counting method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103425967B (en) * | 2013-07-21 | 2016-06-01 | 浙江大学 | A kind of based on stream of people's monitoring method of pedestrian detection and tracking |
| CN103871082A (en) * | 2014-03-31 | 2014-06-18 | 百年金海科技有限公司 | Method for counting people stream based on security and protection video image |
-
2014
- 2014-07-25 CN CN201410360781.5A patent/CN105303191A/en not_active Withdrawn
-
2015
- 2015-01-30 WO PCT/CN2015/072048 patent/WO2015131734A1/en active Application Filing
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101577812A (en) * | 2009-03-06 | 2009-11-11 | 北京中星微电子有限公司 | Method and system for post monitoring |
| CN101872422A (en) * | 2010-02-10 | 2010-10-27 | 杭州海康威视软件有限公司 | People flow rate statistical method and system capable of precisely identifying targets |
| 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 |
| US20120274781A1 (en) * | 2011-04-29 | 2012-11-01 | Siemens Corporation | Marginal space learning for multi-person tracking over mega pixel imagery |
| CN102568005A (en) * | 2011-12-28 | 2012-07-11 | 江苏大学 | Moving object detection method based on Gaussian mixture model |
| CN102799935A (en) * | 2012-06-21 | 2012-11-28 | 武汉烽火众智数字技术有限责任公司 | Human flow counting method based on video analysis technology |
| CN103021059A (en) * | 2012-12-12 | 2013-04-03 | 天津大学 | Video-monitoring-based public transport passenger flow counting method |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107092915A (en) * | 2016-02-18 | 2017-08-25 | 中国移动通信集团浙江有限公司 | A kind of method and apparatus for detecting crowd density |
| CN107092915B (en) * | 2016-02-18 | 2021-03-02 | 中国移动通信集团浙江有限公司 | Method and device for detecting crowd density |
| CN106326851B (en) * | 2016-08-19 | 2019-08-13 | 杭州智诺科技股份有限公司 | A kind of method of number of people detection |
| CN106326851A (en) * | 2016-08-19 | 2017-01-11 | 杭州智诺科技股份有限公司 | Head detection method |
| CN107403137B (en) * | 2017-06-29 | 2020-01-31 | 山东师范大学 | Dense crowd flow calculation method and device based on video |
| CN107403137A (en) * | 2017-06-29 | 2017-11-28 | 山东师范大学 | Dense population method of calculating flux and device based on video |
| CN108197579B (en) * | 2018-01-09 | 2022-05-20 | 杭州智诺科技股份有限公司 | Method for detecting number of people in protection cabin |
| CN108197579A (en) * | 2018-01-09 | 2018-06-22 | 杭州智诺科技股份有限公司 | The detection method of number in protective cabin |
| CN108345842A (en) * | 2018-01-24 | 2018-07-31 | 成都鼎智汇科技有限公司 | A kind of processing method based on big data |
| CN108280427A (en) * | 2018-01-24 | 2018-07-13 | 成都鼎智汇科技有限公司 | A kind of big data processing method based on flow of the people |
| CN108345842B (en) * | 2018-01-24 | 2022-03-04 | 中电长城圣非凡信息系统有限公司 | Big data based processing method |
| CN108280427B (en) * | 2018-01-24 | 2021-11-09 | 广州盖盟达工业品有限公司 | Big data processing method based on pedestrian flow |
| CN110490030A (en) * | 2018-05-15 | 2019-11-22 | 保定市天河电子技术有限公司 | A kind of channel demographic method and system based on radar |
| CN111091529A (en) * | 2018-10-24 | 2020-05-01 | 株式会社理光 | People counting method and system |
| CN111797652A (en) * | 2019-04-09 | 2020-10-20 | 佳能株式会社 | Object tracking method, device and storage medium |
| CN110705408A (en) * | 2019-09-23 | 2020-01-17 | 东南大学 | Indoor people counting method and system based on mixed Gaussian people distribution learning |
| CN112084959A (en) * | 2020-09-11 | 2020-12-15 | 腾讯科技（深圳）有限公司 | Crowd image processing method and device |
| CN112084959B (en) * | 2020-09-11 | 2024-04-16 | 腾讯科技（深圳）有限公司 | Crowd image processing method and device |
| CN112232210A (en) * | 2020-10-16 | 2021-01-15 | 京东方科技集团股份有限公司 | Personnel flow analysis method and system, electronic device and readable storage medium |
| CN112446340A (en) * | 2020-12-07 | 2021-03-05 | 深圳市信义科技有限公司 | Pedestrian searching method and system combining pedestrian local feature and decorative attribute classification and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2015131734A1 (en) | 2015-09-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105303191A (en) | Method and apparatus for counting pedestrians in foresight monitoring scene | |
| CN102542289B (en) | Pedestrian volume statistical method based on plurality of Gaussian counting models | |
| WO2020253308A1 (en) | Human-machine interaction behavior security monitoring and forewarning method for underground belt transportation-related personnel | |
| CN103310444B (en) | A kind of method of the monitoring people counting based on overhead camera head | |
| CN101872418B (en) | Detection method based on group environment abnormal behavior | |
| US20180181803A1 (en) | Pedestrian head identification method and system | |
| CN103605971B (en) | Method and device for capturing face images | |
| CN106203513B (en) | A kind of statistical method based on pedestrian's head and shoulder multi-target detection and tracking | |
| CN102867188B (en) | Method for detecting seat state in meeting place based on cascade structure | |
| Li et al. | Robust people counting in video surveillance: Dataset and system | |
| CN104680557A (en) | Intelligent detection method for abnormal behavior in video sequence image | |
| CN105678811A (en) | Motion-detection-based human body abnormal behavior detection method | |
| CN108229407A (en) | A kind of behavioral value method and system in video analysis | |
| CN103218816A (en) | Crowd density estimation method and pedestrian volume statistical method based on video analysis | |
| CN103605983B (en) | Remnant detection and tracking method | |
| CN102855508B (en) | Opening type campus anti-following system | |
| CN104978567A (en) | Vehicle detection method based on scenario classification | |
| CN102214309A (en) | Special human body recognition method based on head and shoulder model | |
| CN103902966A (en) | Video interaction event analysis method and device base on sequence space-time cube characteristics | |
| CN106127812A (en) | A kind of passenger flow statistical method of non-gate area, passenger station based on video monitoring | |
| KR101472674B1 (en) | Method and apparatus for video surveillance based on detecting abnormal behavior using extraction of trajectories from crowd in images | |
| CN107483894A (en) | Judge to realize the high ferro station video monitoring system of passenger transportation management based on scene | |
| CN104143197A (en) | Detection method for moving vehicles in aerial photography scene | |
| CN113362374A (en) | High-altitude parabolic detection method and system based on target tracking network | |
| CN103632427A (en) | Gate cracking protection method and gate control system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20160203 |