CN106295523A - A kind of public arena based on SVM Pedestrian flow detection method - Google Patents
A kind of public arena based on SVM Pedestrian flow detection method Download PDFInfo
- Publication number
- CN106295523A CN106295523A CN201610615928.XA CN201610615928A CN106295523A CN 106295523 A CN106295523 A CN 106295523A CN 201610615928 A CN201610615928 A CN 201610615928A CN 106295523 A CN106295523 A CN 106295523A
- Authority
- CN
- China
- Prior art keywords
- human body
- svm
- moving target
- feature
- flow detection
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20224—Image subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of public arena based on SVM Pedestrian flow detection method, belong to computer vision field.Its feature includes: video image is stored in original video data storehouse by (1)；(2) utilize background subtraction and frame differential method that the moving target in image is extracted；(3) DPM method is utilized to extract moving target feature；(4) moving target feature being contrasted with the characteristics of human body's model library utilizing SVM to set up, judge whether moving target is human body according to comparing result, whether flow of the people enumerator is added and subtracted；(5) area partition assessor's flow is used when human body shielded area is excessive.The present invention, by DPM feature extraction and SVM training of human body characteristics Sample Storehouse detection human body, has good fault-tolerant ability and calculates speed soon, can be used for densely populated public arena Pedestrian flow detection system sample polytropy.
Description
1. technical field
The invention belongs to computer vision field, particularly relate to machine learning therein and pattern recognition.
2. background technology
Pedestrian flow detection be computer vision field the most popular be also one of the most challenging research direction, have extensively
Application prospect, such as fields such as security protection, traffic, amusements.Existing Pedestrian flow detection system, it is common that by camera collection figure
Picture, is delivered to backstage and processes, and processing procedure is complex, and detection speed is relatively slow, and precision is the highest, is generally not capable of effectively
Solve Pedestrian flow detection problem.
Public place safety problem is more and more paid attention in recent years, and monitoring system is widely used in each public field
Institute.At computer vision field, human detection is to ensure that the basic of Pedestrian flow detection precision, then how to solve real in video
The human detection of existing higher reliability becomes study hotspot.In human detection, most important two links are feature extraction and classification
The design of device.The present invention proposes the Research Thinking of oneself, to a kind of new method of offer to improve the precision of Pedestrian flow detection,
Carry out training of human body Model especially in accordance with SVM and set up characteristics of human body's grader, utilize DPM method that human body is divided into various piece,
Carry out feature identification, reduce false recognition rate.This method can reduce the wrong identification to chaff interference, improves the precision identified.
3. summary of the invention
The present invention is a kind of based on SVM public arena Pedestrian flow detection method designed according to above-mentioned thinking.
The technical scheme is that a kind of public arena based on SVM of offer Pedestrian flow detection method, its feature includes
Following steps:
(1) video image being stored in original video data storehouse, video image includes video file and file attribute；
(2) utilize background subtraction and frame differential method that the moving target in image is extracted；
(3) DPM method is utilized to extract moving target feature；
(4) moving target feature is contrasted with the characteristics of human body's model library utilizing SVM to set up, sentence according to comparing result
Whether disconnected moving target is human body, and whether flow of the people enumerator is added and subtracted；
(5) when human body shielded area is excessive, then area partition assessor's flow is used.
The most further comprising the steps of in described step (1):
A () collects the image of human body standard feature and is stored in master pattern storehouse；
B (), according to master pattern storehouse, according to the characteristics of human body of DPM method identification, utilizes SVM method to human body various piece
Eigenvalue carries out learning training model, forms characteristics of human body's model library.
Described step (2) comprises the following steps:
(2.1) image framing, carries out sub-frame processing to video image to be processed；
(2.2) Image semantic classification, video image to be processed is carried out binaryzation, scale, filtering etc. processes.
The present invention proposes public arena based on SVM Pedestrian flow detection method, mainly includes herein below:
(1) characteristics of human body extracts
This method utilizes DPM method, moving target region in image is carried out Local Features Analysis and extraction, takes
From parts to overall approach, human body is divided into the parts such as head, extremity, trunk, is then respectively trained all parts sample, according to
Geometrical constraint between parts detects whole human body.By recycling DPM after moving target recognition is carried out feature extraction, and
Contrast with characteristics of human body in the feature model library trained in advance, improve the precision of Pedestrian flow detection.
(2) characteristics of human body's model library is set up
The characteristics of human body's model library i.e. foundation of grader, on the basis of according to tradition SVM training pattern method, passes through
DPM feature extracting method, is divided into several part, such as head, extremity and health whole human body, sets up its feature respectively
Storehouse, and then form SVM classifier, construct a human body detector.SVM classifier has the good fault-tolerant energy to sample polytropy
Power and stronger learning capacity, have preferable discrimination and relatively low misclassification rate for different samples, and calculate speed
Hurry up, can quickly carry out substantial amounts of video images detection, can be used for densely populated public arena Pedestrian flow detection system.
4. accompanying drawing explanation
Fig. 1 is the structured flowchart of the system of the application present invention.
5. detailed description of the invention
Below the detailed description of the invention of the present invention is described in further detail.
As it is shown in figure 1, the present invention proposes a kind of Pedestrian flow detection method based on SVM, corresponding system includes: image is pre-
Processing, moving target recognition, DPM feature extraction, human body spy's model library is set up, area reckoning flow of the people.
Image semantic classification includes video image is carried out sub-frame processing, single frames picture carries out binaryzation, processes except hot-tempered etc.；
Moving target recognition is to utilize frame differential method that moving target region is extracted, and then recycles background
Calculus of finite differences is moving object detection out；
DPM feature extraction is the feature utilizing DPM method to extract moving target；
It is to utilize SVM method training human body various piece model and form feature database, at SVM that characteristics of human body's model library is set up
Add by DPM method on the basis of training study, whole human body is divided into several part, such as head, extremity and health, divide
Do not set up feature database, and then form SVM characteristics of human body's grader, set up characteristics of human body's model library；
By the moving target feature detected and the contrast of characteristics of human body's model library, it is determined that the moving target detected is
No for human body, so that flow of the people rolling counters forward；
When the situation of person to person's high superposed, it is impossible to accurately detect number, area reckoning method is used to estimate artificial abortion
Amount.
More than implementing to be only the present invention a kind of embodiment therein, it describes more concrete, but can not therefore manage
Solve as the restriction to the scope of the claims of the present invention.For those skilled in the art, without departing from the inventive concept of the premise, also
Can make some deformation and improvement, these broadly fall into protection scope of the present invention.Therefore, the protection domain of patent of the present invention should
It is as the criterion with claims.
Claims (3)
1. a kind of public arena based on SVM of offer Pedestrian flow detection method is provided, it is characterised in that: its
Including following step:
(1) video image being stored in original video data storehouse, video image includes video file and file attribute；
(2) utilize background subtraction and frame differential method that the moving target in image is extracted；
(3) DPM method is utilized to extract moving target feature；
(4) moving target feature is contrasted with the characteristics of human body's model library utilizing SVM to set up, judge fortune according to comparing result
Whether moving-target is human body, and whether flow of the people enumerator is added and subtracted；
(5) area partition assessor's flow is used when human body shielded area is excessive.
Public arena based on SVM the most according to claim 1 Pedestrian flow detection method, before above-mentioned steps (1) also
Comprise the following steps:
A () collects the image of human body standard feature and is stored in master pattern storehouse；
B (), according to master pattern storehouse, according to the characteristics of human body of DPM method identification, utilizes SVM method to human body various piece feature
Value carries out learning training model, forms characteristics of human body's model library.
Public arena based on SVM the most according to claim 1 Pedestrian flow detection method, the present invention proposes based on SVM
The method of training human body each several part characteristic model, including following components:
(1) characteristics of human body extracts: moving target region in image is carried out feature extraction, takes from parts to overall way
Footpath, is divided into human body the parts such as head, extremity, trunk, is respectively trained each parts sample, detects according to the geometrical constraint between parts
Whole human body.
(2) characteristics of human body's model library is set up: utilize SVM training pattern method, is respectively trained people according to the characteristics of human body that DPM extracts
The component features such as the head of body, extremity, trunk, and then form characteristics of human body's model library.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610615928.XA CN106295523A (en) | 2016-08-01 | 2016-08-01 | A kind of public arena based on SVM Pedestrian flow detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610615928.XA CN106295523A (en) | 2016-08-01 | 2016-08-01 | A kind of public arena based on SVM Pedestrian flow detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106295523A true CN106295523A (en) | 2017-01-04 |
Family
ID=57663438
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610615928.XA Pending CN106295523A (en) | 2016-08-01 | 2016-08-01 | A kind of public arena based on SVM Pedestrian flow detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106295523A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108615020A (en) * | 2018-04-28 | 2018-10-02 | 东莞市华睿电子科技有限公司 | A kind of floating population number statistical method in video monitoring regional |
| CN108647615A (en) * | 2018-04-28 | 2018-10-12 | 东莞市华睿电子科技有限公司 | Method for realizing cinema ticket evasion early warning |
| CN108734148A (en) * | 2018-05-29 | 2018-11-02 | 河南牧业经济学院 | A kind of public arena image information collecting unmanned aerial vehicle control system based on cloud computing |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101847265A (en) * | 2010-04-20 | 2010-09-29 | 上海理工大学 | Method for extracting moving objects and partitioning multiple objects used in bus passenger flow statistical system |
| US20120213426A1 (en) * | 2011-02-22 | 2012-08-23 | The Board Of Trustees Of The Leland Stanford Junior University | Method for Implementing a High-Level Image Representation for Image Analysis |
| CN103679189A (en) * | 2012-09-14 | 2014-03-26 | 华为技术有限公司 | Method and device for recognizing scene |
| CN104933441A (en) * | 2015-06-12 | 2015-09-23 | 北京科富兴科技有限公司 | Target detection system and method |
| CN105138983A (en) * | 2015-08-21 | 2015-12-09 | 燕山大学 | Pedestrian detection method based on weighted part model and selective search segmentation |
| CN105760824A (en) * | 2016-02-02 | 2016-07-13 | 北京进化者机器人科技有限公司 | Moving body tracking method and system |
-
2016
- 2016-08-01 CN CN201610615928.XA patent/CN106295523A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101847265A (en) * | 2010-04-20 | 2010-09-29 | 上海理工大学 | Method for extracting moving objects and partitioning multiple objects used in bus passenger flow statistical system |
| US20120213426A1 (en) * | 2011-02-22 | 2012-08-23 | The Board Of Trustees Of The Leland Stanford Junior University | Method for Implementing a High-Level Image Representation for Image Analysis |
| CN103679189A (en) * | 2012-09-14 | 2014-03-26 | 华为技术有限公司 | Method and device for recognizing scene |
| CN104933441A (en) * | 2015-06-12 | 2015-09-23 | 北京科富兴科技有限公司 | Target detection system and method |
| CN105138983A (en) * | 2015-08-21 | 2015-12-09 | 燕山大学 | Pedestrian detection method based on weighted part model and selective search segmentation |
| CN105760824A (en) * | 2016-02-02 | 2016-07-13 | 北京进化者机器人科技有限公司 | Moving body tracking method and system |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108615020A (en) * | 2018-04-28 | 2018-10-02 | 东莞市华睿电子科技有限公司 | A kind of floating population number statistical method in video monitoring regional |
| CN108647615A (en) * | 2018-04-28 | 2018-10-12 | 东莞市华睿电子科技有限公司 | Method for realizing cinema ticket evasion early warning |
| CN108734148A (en) * | 2018-05-29 | 2018-11-02 | 河南牧业经济学院 | A kind of public arena image information collecting unmanned aerial vehicle control system based on cloud computing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jiang et al. | Automatic soccer video event detection based on a deep neural network combined cnn and rnn | |
| JP6549797B2 (en) | Method and system for identifying head of passerby | |
| CN104156942B (en) | Detection method for remnants in complex environment | |
| CN105095856B (en) | Face identification method is blocked based on mask | |
| CN101477626B (en) | Method for detecting human head and shoulder in video of complicated scene | |
| CN105187785B (en) | A kind of across bayonet pedestrian's identifying system and method based on choice of dynamical notable feature | |
| WO2015131734A1 (en) | Method, device, and storage medium for pedestrian counting in forward looking surveillance scenario | |
| CN107943837A (en) | A kind of video abstraction generating method of foreground target key frame | |
| CN107622258A (en) | A Fast Pedestrian Detection Method Combining Static Low-level Features and Motion Information | |
| US20120093362A1 (en) | Device and method for detecting specific object in sequence of images and video camera device | |
| CN103778409A (en) | Human face identification method based on human face characteristic data mining and device | |
| Rodriguez et al. | Detecting and segmenting humans in crowded scenes | |
| CN102930553A (en) | Method and device for identifying objectionable video content | |
| CN107909027A (en) | It is a kind of that there is the quick human body target detection method for blocking processing | |
| CN102768726B (en) | Pedestrian detection method for preventing pedestrian collision | |
| CN107067413A (en) | A kind of moving target detecting method of time-space domain statistical match local feature | |
| CN105554462B (en) | A kind of remnant object detection method | |
| WO2018040306A1 (en) | Method for detecting frequent passers-by in monitoring video | |
| CN105787472A (en) | Abnormal behavior detection method based on time-space Laplacian Eigenmaps learning | |
| CN109784130A (en) | Pedestrian recognition methods and its device and equipment again | |
| CN103310444A (en) | Method of monitoring pedestrians and counting based on overhead camera | |
| CN106919921A (en) | With reference to sub-space learning and the gait recognition method and system of tensor neutral net | |
| CN110991397A (en) | Traveling direction determining method and related equipment | |
| CN103400155A (en) | Pornographic video detection method based on semi-supervised learning of images | |
| CN109034247A (en) | A kind of recognition of face sample extraction method of the higher degree based on track algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170104 |
|
| WD01 | Invention patent application deemed withdrawn after publication |