CN103839085A - Train carriage abnormal crowd density detection method - Google Patents
Train carriage abnormal crowd density detection method Download PDFInfo
- Publication number
- CN103839085A CN103839085A CN201410094075.0A CN201410094075A CN103839085A CN 103839085 A CN103839085 A CN 103839085A CN 201410094075 A CN201410094075 A CN 201410094075A CN 103839085 A CN103839085 A CN 103839085A
- Authority
- CN
- China
- Prior art keywords
- crowd density
- compartment
- feature
- density
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 53
- 238000001514 detection method Methods 0.000 title claims abstract description 15
- 238000000034 method Methods 0.000 claims abstract description 42
- 230000004927 fusion Effects 0.000 claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 6
- 239000013598 vector Substances 0.000 claims description 26
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 7
- 238000007637 random forest analysis Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 abstract description 5
- 238000005286 illumination Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000003672 processing method Methods 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000003203 everyday effect Effects 0.000 description 2
- 230000002265 prevention Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a train carriage abnormal crowd density detection method. The method includes the following steps that a plurality of carriage sample images with different crowd density grades are collected, and multimodal fusion characteristics of the images are obtained; a crowd density classifier is obtained through training; multimodal fusion characteristics of an image to be detected are obtained; according to the crowd density classifier, the crowd density grade of a carriage corresponding to the image to be detected is obtained, and accordingly whether the crowd density in the carriage is abnormal or not is judged; abnormal information of related crowd density is recorded automatically. According to the method, the multimodal fusion characteristics are used, abnormal crowd density scenes are learned and identified automatically, and therefore abnormal crowd density can be identified and recorded automatically in a real-time mode in the running process of a train. The method is insensitive to train scene crowd shielding, illumination and slight deformation of a camera, and is suitable for train abnormal crowd density detection through 360 cameras or bolt cameras.
Description
Technical field
The invention belongs to technical field of video processing, relate in particular to one according to railway car video, in real time, automatically analyze the method that whether has abnormal crowd density.
Background technology
At present, domestic nearly all subway train has all been installed video frequency graphic monitoring system.Train is in the time of operation, and video monitoring system automatically records the situation in compartment and stores relevant video.The situation of train supervision is at present: subway train volume of the flow of passengers every day differing greatly in each period, railway car relative closure, environment relative complex in railway car, video image acquisition is limited, illumination is very fast according to changes in environmental conditions, camera has 360 cameras and gunlock camera, view data memory space is large, the video data volume that train was runed 3 hours in one day exceedes 10Gbit, often many train operations simultaneously of city or area, more piece compartment shares a set of video image system, and the video data volume that need to store every day is very large.Once have an accident, need to manually transfer with artificial enquiry store video in whether there is abnormality, very labor intensive and material resources.
International terrorist incident happens occasionally, and people's awareness of safety promotes day by day, and promoting public transport safety and anti-terrorism is also a kind of common recognition at home.Identifying train unusual condition how in real time, is accurately fast processing train accident, the important step that ensures public safety, is also the active demand of modern Intelligentized subway train.And it is abnormal to detect in real time railway car crowd density, be that compartment personnel dredge, improve the primary demand of the Train Managements such as program comfort level and prevention group gather around.The fast development of modern video image processing techniques, the development of the technology such as especially image processing, computer vision and artificial intelligence, can be achieved real-time analysis compartment abnormality.Although in recent years, progressively increase for the image processing method of automatically identifying crowd density, also there is no the abnormal crowd density in a kind of suitable image processing method energy automatic decision railway car at present.
Conventional crowd density determination methods mainly contains two large classes at present: based on pedestrian or human body parts (number of people, above the waist etc.) detection and the method based on statistical learning.
Based on the method for pedestrian or human body parts (number of people, etc.) detection, need to from image, see that pedestrian, the number of people or the upper part of the body etc. have health or the body part of notable feature above the waist.And under railway car condition, crowd's serious shielding often, even if want to see that the complete number of people can not guarantee, therefore whether extremely comparatively these class methods for judging the crowd density difficulty of railway car.
Based on the method for statistical learning, there is at present the crowd density in subway detection method based on video, be mainly by Gaussian Background modeling method, extraction prospect, estimates crowd density according to the area of prospect; When train operation, compartment background complexity, illumination variation, the irregular and serious shielding condition of compartment passenger moving, therefore the effect of Gauss's modeling is bad, and the accuracy rate of identification crowd density is lower.
Based on the method for statistical learning, the current dynamic texture based on time-space domain local binary pattern in addition and the crowd density Methods For Global Estimation of support vector machine, the method has certain effect for analyst's population density, has some application in public in the estimation of crowd density.But because single textural characteristics reaches middle-high density when above at crowd density, the ability of distinguishing crowd density significantly declines, judging nicety rate is on the low side and need very large training sample quantity could obtain suitable effect.And railway car may be within upper and lower class or certain period, crowd density is for a long time in middle-high density, is now just difficult to detect wherein abnormal.
Summary of the invention
In order to overcome above-mentioned the deficiencies in the prior art, the invention provides the detection method of the abnormal crowd density of a kind of railway car.The method shares the present situation of a set of video image system according to more piece railway car under metro environment, estimates in railway car, whether there is abnormal crowd density according to the density variation (density rating difference is greater than 1 grade) of two joints or more piece railway car.
The detection method of the abnormal crowd density of a kind of railway car provided by the invention comprises the following steps:
Step 2, extract described sample image textural characteristics separately, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature;
The textural characteristics of a certain sample image obtaining is extracted in step 3, fusion, Surf, Fast, Harris unique point feature, and foreground image Area Ratio feature and light stream density feature, generate multi-modal fusion feature;
Step 4, according to the multi-modal fusion feature of described several sample images, training obtains crowd density sorter;
Step 5, from the monitor video of railway car, intercept image to be detected, extract successively the textural characteristics of described image to be detected, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature, and these features are merged, obtain the multi-modal fusion feature of described image to be detected;
Step 6, the multi-modal fusion feature of described image to be detected is inputed to described crowd density sorter, obtain the crowd density grade in the corresponding compartment of described image to be detected;
The crowd density grade in step 7, this compartment of obtaining according to described step 6 judges that whether the crowd density in this compartment is abnormal;
The crowd density abnormal information in step 8, the abnormal compartment of automatic recorder's population density.
The advantage of the technical solution adopted in the present invention has:
1, the present invention shares a set of video image system present situation according to more piece compartment under metro environment, estimates in railway car, whether there is abnormal crowd density according to two joints or the difference (density rating difference is greater than 1 grade) of more piece compartment density;
2, the present invention uses the method based on statistical machine learning, replace exist at present extensively manually check video method, can realize and in real time, automatically detect subway train compartment crowd density and whether exist extremely, and coach number and the time of origin in the record generation abnormal compartment of crowd density, and the corresponding abnormal picture of crowd density;
3, the present invention has introduced Surf, Fast, Harris unique point and light stream density feature, has improved existing single utilization foreground image or textural characteristics and has detected subway train compartment crowd density, the problem that accuracy of detection is not high;
4, the present invention is LBP texture feature vector, Surf, Fast, Harris unique point quantity, foreground image Area Ratio, light stream density feature merges mutually, generate multi-modal fusion proper vector, make various features acting in conjunction in distinguishing crowd density, improve the accuracy that detects crowd density;
5, the present invention adopts efficient random forest learning algorithm, and from the railway car image pattern learning of different density ratings to random forest sorter, and the sorter that study is arrived is for real-time crowd density grade separation.
Compared with prior art, beneficial effect of the present invention has:
1, the present invention provides a kind of in the time that subway train moves first, in real time, automatically detect the abnormal method of crowd density, can on video monitoring system basis, existing subway train compartment, suitably adjust, to realize in real time, automatically detecting and record of the abnormal crowd density in subway train compartment;
2, for also there is no at present a kind of suitable abnormal problem of image processing method automatic decision subway train compartment crowd density, the present invention is in conjunction with subway railway car environment actual state, the method that has proposed the abnormal crowd density of a kind of two joints or more piece subway train compartment crowd density grade difference (density rating difference is greater than 1 grade) identification compartment, has reduced and has only relied on the high wrong report situation that judges that crowd density is abnormal of crowd density.The present invention can help subway managerial personnel to dredge passenger's (guiding high density compartment passenger transfer to low-density compartment), and prevention passenger is crowded, trample, and improves comfort of passenger and satisfaction;
3, the invention provides a kind of method of effective integration foreground image feature, textural characteristics, Surf, Fast, Harris unique point and light stream density feature, highlight crowd density feature, and make several features can acting in conjunction under subway train vehicle environment, improve the accuracy rate that identification crowd density detects.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of the abnormal crowd density detection method of a kind of railway car that proposes of the present invention.
Fig. 2 is that the process flow diagram of abnormal crowd density estimation is combined in more piece compartment according to an embodiment of the invention.
Embodiment
For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with specific embodiment, and with reference to accompanying drawing, the present invention is described in more detail.
The present invention gathers the sample image of the railway car of different densities grade in advance, calculate the textural characteristics of sample image, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature, and generate multi-modal fusion feature based on above-mentioned multi-modal feature; Multi-modal fusion features training based on obtaining obtains crowd density sorter; In the time of train operation, from monitor video, intercept and obtain detected image, calculate equally the multi-modal fusion feature of detected image; According to described crowd density sorter, the multi-modal fusion feature of described detected image is classified; judge the crowd density grade in the corresponding compartment of described detected image; and then estimate whether this compartment exists crowd density abnormal; and abnormal coach number and the time of origin of crowd density occurs automatic record simultaneously, intercept and the now abnormal picture of crowd density of correspondence of storage.
Fig. 1 is the process flow diagram of the abnormal crowd density detection method of a kind of railway car that proposes of the present invention, as shown in Figure 1, said method comprising the steps of:
Described crowd density grade comprises low crowd density, medium crowd density, high crowd density and superelevation crowd density, for convenience's sake, low crowd density can be labeled as to density rating 0; Medium crowd density is labeled as to density rating 1; High crowd density is labeled as to density rating 2; Superelevation crowd density is labeled as to density rating 3.
Step 2, extract described sample image textural characteristics separately, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature;
How texture feature extraction, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature are described below simply.
1) textural characteristics
In an embodiment of the present invention, utilize the LBP operator of multiple subregions in image to extract the textural characteristics of described image.Described LBP operator definitions is for being conventionally taken as 3 × 3 at m × m() window in, take the pixel value of the central pixel point of window as threshold value, the pixel value of m × m-1 the pixel adjacent with this central pixel point is compared with it respectively, if the pixel value of surrounding pixel point is greater than the pixel value of central pixel point, the position of this central pixel point is marked as 1, otherwise be labeled as 0, like this, through with m × m neighborhood in m × m-1 point relatively after, can produce the binary number of a m × m-1 position corresponding with this central pixel point, it is LBP code that this binary number is converted to decimal number conventionally, totally 256 kinds, obtain the LBP value of this window center pixel, this LBP value just can be used for reflecting the texture information of this window area.
Particularly, the step of the LBP textural characteristics of extraction piece image comprises the following steps:
Step 211, described image is divided into multiple n × n(wherein, n>m, more preferably, n>12, such as being taken as 16) subregion;
Step 212, for the some pixels in every sub regions, by its pixel value respectively with its m × m(3 × 3) m × m-1(8 in neighborhood) pixel value of individual pixel compares, if the pixel value of surrounding pixel point is greater than the pixel value of this pixel, the position of this pixel is marked as 1, otherwise be labeled as 0, like this, through with m × m(3 × 3) m × m-1(8 in neighborhood) and individual point relatively after, can produce a m × m-1(8 corresponding with this pixel) position binary number, this binary number is exactly the LBP value that this pixel is corresponding;
Step 213, calculate the statistic histogram of respective sub-areas according to the LBP value of every sub regions pixel, be the frequency of LBP value (supposition the is decimal numeral LBP value) appearance of each pixel in respective sub-areas, and the statistic histogram obtaining is normalized;
Step 214, the statistic histogram after the normalization of every sub regions is connected, obtain a proper vector, as the LBP textural characteristics of described image.
Step 215, according to maximal value and minimum value in described LBP texture feature vector, described LBP texture feature vector is divided into p interval, wherein, the value of p can be determined and adjust according to the needs of practical application, such as can be taken as 9, each element of adding up in described LBP texture feature vector drops on the frequency in each interval, thereby obtains the proper vector of a p dimension, be normalized again, obtained the LBP textural characteristics of p dimension.
Certain described textural characteristics, except using described LBP textural characteristics, can also use the gray level co-occurrence matrixes feature of the proper vector that is normalized to certain dimension and other features of texture that can presentation video.
2) Surf, Fast, Harris unique point feature
In this step, can extract by several different methods of the prior art the extraction of Surf, Fast, Harris unique point, such as can be according to the Surf feature of Opencv built-in function synthetic image, Fast and Harris feature.
In an embodiment of the present invention, for the convenience of calculating, also by Surf unique point quantity, Fast unique point quantity and Harris unique point quantity are normalized, such as the image that is 704*576 for size, can be normalized with 1000, generate the 3 dimensional feature point proper vectors that formed by above-mentioned 3 unique point quantity.
3) foreground image Area Ratio feature
In described step 2, the extraction of described foreground image Area Ratio feature comprises the following steps:
Step 221, store the empty wagons railway carriage or compartment image image as a setting of each railway car;
Step 222, by present image subtracting background image, obtain corresponding foreground image;
Step 223, to described foreground image through gaussian filtering (such as the gaussian filtering of 3*3), obtain described foreground image UNICOM region boundary rectangle area sum, the total pixel by it divided by described present image, obtains the foreground image Area Ratio feature of 1 dimension.
4) light stream density feature
The region that crowd density is larger, the light stream density between adjacent two two field pictures is just larger; Otherwise light stream density is just less.In this step, can extract described light stream density feature by several different methods of the prior art, such as getting the previous frame image of present image and storage, for the surf unique point of extracting before, use KLT(Kanade-Lucas-Tomasi of the prior art) algorithm, calculate the quantity of the surf unique point that is subjected to displacement variation, be light stream numerical value, be normalized, such as divided by 1000, just obtained the light stream density feature of 1 dimension.
The textural characteristics of a certain sample image obtaining is extracted in step 3, fusion, Surf, Fast, Harris unique point feature, and foreground image Area Ratio feature and light stream density feature, generate multi-modal fusion feature;
In an embodiment of the present invention, can adopt the method that combines successively texture feature vector, unique point proper vector, foreground image Area Ratio proper vector and light stream density feature vector, obtain described multi-modal fusion feature, certainly also can adopt other method, such as the mode such as weight that suitably increases or reduce dimension, the increase of each proper vector or reduce each proper vector obtains described multi-modal fusion feature.
Step 4, according to the multi-modal fusion feature of described several sample images, training obtains crowd density sorter;
In this step, can obtain described crowd density sorter by several different methods of the prior art, such as random forest learning algorithm or support vector machine learning algorithm.
Step 5, from the monitor video of railway car, intercept image to be detected, extract successively the textural characteristics of described image to be detected, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature, and these features are merged, obtain the multi-modal fusion feature of described image to be detected;
In this step, extract the textural characteristics of image to be detected, Surf, Fast, Harris unique point feature, the step of foreground image Area Ratio feature and light stream density feature, and these features are merged, the step that obtains multi-modal fusion feature is all similar with described step 2,3 description, does not repeat them here.
Step 6, the multi-modal fusion feature of described image to be detected is inputed to described crowd density sorter, obtain the crowd density grade in the corresponding compartment of described image to be detected;
The crowd density grade in step 7, this compartment of obtaining according to described step 6 judges that whether the crowd density in this compartment is abnormal;
In this step, for the less demanding situation of wrong report, can be directly that the compartment of superelevation crowd density is estimated as the compartment that exists crowd density abnormal by crowd density grade; The situation of having relatively high expectations for wrong report, can judge that whether the crowd density in corresponding compartment is abnormal according to the joint density level status in two joints or more piece compartment, when there is notable difference in the crowd density grade in two joints or more piece compartment even, compartment high crowd density grade is estimated as to the abnormal compartment of crowd density, compartment low crowd density grade is estimated as to the normal compartment of crowd density.
Judge that according to joint density level status whether abnormal the crowd density in corresponding compartment step be further comprising the steps:
Step 71, K is saved to compartment as one group of joint-detection object, wherein, the value of K can be set according to the needs of practical application, and in an embodiment of the present invention, the value of K is 2 or 3;
Step 72, according to the crowd density grade in described step 6 successively calculating K joint compartment: k1, k2 ..., kK;
If have at least 1 crowd density grade to equal 3 in the crowd density grade in step 73 K joint compartment, be judged as superelevation crowd density (highest crowd density), and the crowd density rank difference of adjacent compartment is no more than 1, | k1-k2|>1, | k2-k3|>1 ..., | kK-k1|>1 has 1 generation at least, and the compartment that is 3 by density rating is estimated as the abnormal compartment of crowd density.
Take three joint compartments as example, the process flow diagram that abnormal crowd density estimation is combined in more piece compartment as shown in Figure 2.
The crowd density abnormal information in step 8, the abnormal compartment of automatic recorder's population density.
In the time that described step 7 detects that the crowd density in a certain compartment is abnormal; also can automatically record abnormal coach number and the time of origin of crowd density occurs; intercept and the now abnormal picture of crowd density of correspondence of storage, and the information of record is bound together, so that later checking and downloading.
Above-described specific embodiment; object of the present invention, technical scheme and beneficial effect are further described; institute is understood that; the foregoing is only specific embodiments of the invention; be not limited to the present invention; within the spirit and principles in the present invention all, any modification of making, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.
Claims (10)
1. a detection method for the abnormal crowd density of railway car, is characterized in that, the method comprises the following steps:
Step 1, collection and storage have several railway car sample images of different crowd density rating, and are the corresponding crowd density grade of described sample image mark;
Step 2, extract described sample image textural characteristics separately, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature;
The textural characteristics of a certain sample image obtaining is extracted in step 3, fusion, Surf, Fast, Harris unique point feature, and foreground image Area Ratio feature and light stream density feature, generate multi-modal fusion feature;
Step 4, according to the multi-modal fusion feature of described several sample images, training obtains crowd density sorter;
Step 5, from the monitor video of railway car, intercept image to be detected, extract successively the textural characteristics of described image to be detected, Surf, Fast, Harris unique point feature, foreground image Area Ratio feature and light stream density feature, and these features are merged, obtain the multi-modal fusion feature of described image to be detected;
Step 6, the multi-modal fusion feature of described image to be detected is inputed to described crowd density sorter, obtain the crowd density grade in the corresponding compartment of described image to be detected;
The crowd density grade in step 7, this compartment of obtaining according to described step 6 judges that whether the crowd density in this compartment is abnormal;
The crowd density abnormal information in step 8, the abnormal compartment of automatic recorder's population density.
2. method according to claim 1, is characterized in that, described crowd density grade comprises low crowd density, medium crowd density, high crowd density and superelevation crowd density.
3. method according to claim 1, is characterized in that, described textural characteristics is the gray level co-occurrence matrixes feature of LBP textural characteristics or the proper vector that is normalized to certain dimension.
4. method according to claim 3, is characterized in that, in the time that described textural characteristics is LBP textural characteristics, the step of extracting described LBP textural characteristics comprises the following steps:
Step 211, described image is divided into the subregion of multiple n × n;
Step 212, for the some pixels in every sub regions, by its pixel value respectively with its 3 × 3 neighborhood in the pixel value of 8 pixels compare, if the pixel value of surrounding pixel point is greater than the pixel value of this pixel, the position of this pixel is marked as 1, otherwise be labeled as 0, through with 3 × 3 neighborhoods in 8 points relatively after, obtain the binary number of 8 corresponding with this pixel, the LBP value that this pixel is corresponding;
Step 213, calculate the statistic histogram of respective sub-areas according to the LBP value of every sub regions pixel, and the statistic histogram obtaining is normalized;
Step 214, the statistic histogram after the normalization of every sub regions is connected, obtain the LBP textural characteristics of described image;
Step 215, according to maximal value and minimum value in described LBP texture feature vector, described LBP texture feature vector is divided into p interval, add up each element in described LBP texture feature vector and drop on the frequency in each interval, obtain the proper vector of a p dimension, be normalized, obtained the LBP textural characteristics of p dimension.
5. method according to claim 1, is characterized in that, in described step 2, the extraction of described foreground image Area Ratio feature comprises the following steps:
Step 221, store the empty wagons railway carriage or compartment image image as a setting of each railway car;
Step 222, by present image subtracting background image, obtain corresponding foreground image;
Step 223, to described foreground image through gaussian filtering, obtain described foreground image UNICOM region boundary rectangle area sum, the total pixel by it divided by described present image, obtains described foreground image Area Ratio feature.
6. method according to claim 1, is characterized in that, in described step 2, the extraction of described light stream density feature comprises the following steps:
Get the previous frame image of present image and storage;
For the surf unique point of extracting before, use KLT algorithm to calculate the quantity of the surf unique point that is subjected to displacement variation;
This quantity is normalized, obtains described light stream density feature.
7. method according to claim 1, it is characterized in that, while generating multi-modal fusion feature in described step 3, adopt the method that combines successively texture feature vector, unique point proper vector, foreground image Area Ratio proper vector and light stream density feature vector, or increase/reduce the dimension of each proper vector, or the weight that increases/reduce each proper vector obtains described multi-modal fusion feature.
8. method according to claim 1, is characterized in that, adopts random forest learning algorithm or support vector machine learning algorithm to train and obtain described crowd density sorter in described step 4.
9. method according to claim 1, is characterized in that, in described step 7, the compartment that is directly superelevation crowd density by crowd density grade is estimated as the compartment that exists crowd density abnormal; Or whether the crowd density that judges corresponding compartment according to the joint density level status in two joints or more piece compartment is abnormal, when there is notable difference in the crowd density grade in two joints or more piece compartment even, compartment high crowd density grade is estimated as to the abnormal compartment of crowd density, compartment low crowd density grade is estimated as to the normal compartment of crowd density.
10. method according to claim 9, is characterized in that, if described step 7 judges that according to joint density level status whether the crowd density in corresponding compartment is abnormal, the step of this judgement is further comprising the steps:
Step 71, K is saved to compartment as one group of joint-detection object;
Step 72, according to the crowd density grade in described step 6 successively calculating K joint compartment;
Be highest crowd density if having 1 crowd density grade at least in the crowd density grade in step 73 K joint compartment, and the crowd density rank difference of adjacent compartment is no more than 1, the compartment of highest crowd density grade is estimated as to the abnormal compartment of crowd density.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410094075.0A CN103839085B (en) | 2014-03-14 | 2014-03-14 | A kind of detection method of compartment exception crowd density |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410094075.0A CN103839085B (en) | 2014-03-14 | 2014-03-14 | A kind of detection method of compartment exception crowd density |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103839085A true CN103839085A (en) | 2014-06-04 |
| CN103839085B CN103839085B (en) | 2018-06-19 |
Family
ID=50802563
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410094075.0A Active CN103839085B (en) | 2014-03-14 | 2014-03-14 | A kind of detection method of compartment exception crowd density |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103839085B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268898A (en) * | 2014-09-15 | 2015-01-07 | 郑州天迈科技股份有限公司 | Method for detecting density of passengers in bus on basis of image analysis |
| CN105184245A (en) * | 2015-08-28 | 2015-12-23 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | Multi-characteristic fusion population density estimation method |
| CN106033548A (en) * | 2015-03-13 | 2016-10-19 | 中国科学院西安光学精密机械研究所 | Crowd abnormity detection method based on improved dictionary learning |
| CN106651878A (en) * | 2016-12-21 | 2017-05-10 | 福建师范大学 | Method for extracting straight line from local invariant feature points |
| CN107016696A (en) * | 2017-03-31 | 2017-08-04 | 广州地理研究所 | A kind of passenger flow density detection method and device |
| CN107295318A (en) * | 2017-08-23 | 2017-10-24 | 无锡北斗星通信息科技有限公司 | Colour projection's platform based on image procossing |
| CN107371004A (en) * | 2017-08-23 | 2017-11-21 | 无锡北斗星通信息科技有限公司 | A kind of method of colour image projection |
| CN109086801A (en) * | 2018-07-06 | 2018-12-25 | 湖北工业大学 | A kind of image classification method based on improvement LBP feature extraction |
| CN109117791A (en) * | 2018-08-14 | 2019-01-01 | 中国电子科技集团公司第三十八研究所 | A kind of crowd density drawing generating method based on expansion convolution |
| CN109919066A (en) * | 2019-02-27 | 2019-06-21 | 湖南信达通信息技术有限公司 | The method and apparatus of passenger's density anomaly in a kind of detection rail transit cars |
| CN111460246A (en) * | 2019-12-19 | 2020-07-28 | 南京柏跃软件有限公司 | Real-time activity abnormal person discovery method based on data mining and density detection |
| CN112990272A (en) * | 2021-02-19 | 2021-06-18 | 上海理工大学 | Sensor optimization selection method for fault diagnosis of water chilling unit |
| CN113158854A (en) * | 2021-04-08 | 2021-07-23 | 东北大学秦皇岛分校 | Automatic monitoring train safety operation method based on multi-mode information fusion |
| CN113538401A (en) * | 2021-07-29 | 2021-10-22 | 燕山大学 | Crowd counting method and system combining cross-modal information in complex scene |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093198A (en) * | 2013-01-15 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Crowd density monitoring method and device |
| CN103218816A (en) * | 2013-04-18 | 2013-07-24 | 中山大学 | Crowd density estimation method and pedestrian volume statistical method based on video analysis |
-
2014
- 2014-03-14 CN CN201410094075.0A patent/CN103839085B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093198A (en) * | 2013-01-15 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Crowd density monitoring method and device |
| CN103218816A (en) * | 2013-04-18 | 2013-07-24 | 中山大学 | Crowd density estimation method and pedestrian volume statistical method based on video analysis |
Non-Patent Citations (1)
| Title |
|---|
| NAN DONG ET AL.: "Crowd Density Estimation Using Sparse Texture Features", 《JOURNAL OF CONVERGENCE INFORMATION TECHNOLOGY》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268898A (en) * | 2014-09-15 | 2015-01-07 | 郑州天迈科技股份有限公司 | Method for detecting density of passengers in bus on basis of image analysis |
| CN106033548A (en) * | 2015-03-13 | 2016-10-19 | 中国科学院西安光学精密机械研究所 | Crowd abnormity detection method based on improved dictionary learning |
| CN106033548B (en) * | 2015-03-13 | 2021-04-20 | 中国科学院西安光学精密机械研究所 | Crowd abnormity detection method based on improved dictionary learning |
| CN105184245A (en) * | 2015-08-28 | 2015-12-23 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | Multi-characteristic fusion population density estimation method |
| CN105184245B (en) * | 2015-08-28 | 2018-12-21 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of crowd density estimation method of multiple features fusion |
| CN106651878B (en) * | 2016-12-21 | 2019-06-11 | 福建师范大学 | A method of for extracting straight line from local invariant feature point |
| CN106651878A (en) * | 2016-12-21 | 2017-05-10 | 福建师范大学 | Method for extracting straight line from local invariant feature points |
| CN107016696A (en) * | 2017-03-31 | 2017-08-04 | 广州地理研究所 | A kind of passenger flow density detection method and device |
| CN107295318A (en) * | 2017-08-23 | 2017-10-24 | 无锡北斗星通信息科技有限公司 | Colour projection's platform based on image procossing |
| CN107371004A (en) * | 2017-08-23 | 2017-11-21 | 无锡北斗星通信息科技有限公司 | A kind of method of colour image projection |
| CN109086801A (en) * | 2018-07-06 | 2018-12-25 | 湖北工业大学 | A kind of image classification method based on improvement LBP feature extraction |
| CN109117791A (en) * | 2018-08-14 | 2019-01-01 | 中国电子科技集团公司第三十八研究所 | A kind of crowd density drawing generating method based on expansion convolution |
| CN109919066A (en) * | 2019-02-27 | 2019-06-21 | 湖南信达通信息技术有限公司 | The method and apparatus of passenger's density anomaly in a kind of detection rail transit cars |
| CN109919066B (en) * | 2019-02-27 | 2021-05-25 | 湖南信达通信息技术有限公司 | Method and device for detecting density abnormality of passengers in rail transit carriage |
| CN111460246A (en) * | 2019-12-19 | 2020-07-28 | 南京柏跃软件有限公司 | Real-time activity abnormal person discovery method based on data mining and density detection |
| CN111460246B (en) * | 2019-12-19 | 2020-12-08 | 南京柏跃软件有限公司 | Real-time activity abnormal person discovery method based on data mining and density detection |
| CN112990272A (en) * | 2021-02-19 | 2021-06-18 | 上海理工大学 | Sensor optimization selection method for fault diagnosis of water chilling unit |
| CN113158854A (en) * | 2021-04-08 | 2021-07-23 | 东北大学秦皇岛分校 | Automatic monitoring train safety operation method based on multi-mode information fusion |
| CN113158854B (en) * | 2021-04-08 | 2022-03-22 | 东北大学秦皇岛分校 | Automatic monitoring train safety operation method based on multi-mode information fusion |
| CN113538401A (en) * | 2021-07-29 | 2021-10-22 | 燕山大学 | Crowd counting method and system combining cross-modal information in complex scene |
| CN113538401B (en) * | 2021-07-29 | 2022-04-05 | 燕山大学 | Crowd counting method and system combining cross-modal information in complex scene |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103839085B (en) | 2018-06-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103839085A (en) | Train carriage abnormal crowd density detection method | |
| CN110956094B (en) | RGB-D multi-mode fusion personnel detection method based on asymmetric double-flow network | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| CN102348128B (en) | Surveillance camera system having camera malfunction detection function | |
| CN104091171B (en) | Vehicle-mounted far infrared pedestrian detecting system and method based on local feature | |
| CN102867188B (en) | Method for detecting seat state in meeting place based on cascade structure | |
| CN105187785B (en) | A kind of across bayonet pedestrian's identifying system and method based on choice of dynamical notable feature | |
| CN102903124A (en) | Moving object detection method | |
| CN106339657B (en) | Crop straw burning monitoring method based on monitor video, device | |
| CN110852190B (en) | Driving behavior recognition method and system integrating target detection and gesture recognition | |
| KR20150029006A (en) | Unsupervised learning of feature anomalies for a video surveillance system | |
| CN105095905A (en) | Target recognition method and target recognition device | |
| CN103473570A (en) | Self-adaptive video scene pedestrian detection method | |
| CN111723773B (en) | Method and device for detecting carryover, electronic equipment and readable storage medium | |
| CN111738342A (en) | Pantograph foreign matter detection method, storage medium and computer equipment | |
| CN108804987B (en) | Door opening and closing state detection method and device and people flow detection system | |
| Karpagavalli et al. | Estimating the density of the people and counting the number of people in a crowd environment for human safety | |
| CN105005773A (en) | Pedestrian detection method with integration of time domain information and spatial domain information | |
| CN107092878A (en) | It is a kind of based on hybrid classifer can autonomous learning multi-target detection method | |
| CN109359593A (en) | A kind of sleet environment fuzzy pictures monitoring and pre-alarming method based on image local grid | |
| CN100546380C (en) | Target detection and tracking at night based on visual characteristic | |
| CN105095891A (en) | Human face capturing method, device and system | |
| CN112508893B (en) | Method and system for detecting tiny foreign matters between double rails of railway based on machine vision | |
| Panchenko et al. | Improvement of the accuracy of determining movement parameters of cuts on classification humps by methods of video analysis | |
| CN113343926A (en) | Driver fatigue detection method based on convolutional neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |