CN103714697B - A kind of method of recognition and tracking criminal vehicle - Google Patents
A kind of method of recognition and tracking criminal vehicle Download PDFInfo
- Publication number
- CN103714697B CN103714697B CN201310714749.8A CN201310714749A CN103714697B CN 103714697 B CN103714697 B CN 103714697B CN 201310714749 A CN201310714749 A CN 201310714749A CN 103714697 B CN103714697 B CN 103714697B
- Authority
- CN
- China
- Prior art keywords
- criminal
- vehicle
- feature
- recognition
- tracking
- 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.)
- Active
Links
Abstract
The invention discloses a kind of method of recognition and tracking criminal vehicle, comprise the following steps: (1) obtains criminal's information of vehicles by alarm people or monitoring video, determine target vehicle, (2) its Invariance feature is extracted according to the image information of target vehicle, as surf feature, (3) quantize to be transferred to CCC after surf feature becomes profiler, the potential region of criminal is determined according to guilty place and time etc., (4) criminal's vehicle characteristics describer divides and reaches each scouting site by CCC, scope of reconnaissance is reduced after coupling scout vehicle, (5) CCC sends profiler to the front cruiser in scope of reconnaissance, (6) surf characteristic matching is carried out to monitoring video real-time in cruiser, the match is successful for front vehicles, then perform tracking.
Description
Technical field
The present invention relates to intelligent traffic monitoring technical field, particularly a kind of method of recognition and tracking criminal vehicle.
Background technology
Along with the development of society, the living standard of people improves constantly, but the sense of security of people may not be certain to increase, the crime rate of China is growing on and in recent years, the criminal offences such as theft, robbery, smuggling have had a strong impact on the safety of life and property of the public, and vehicle then becomes the main escape instrument of these criminals.Even if police has locked criminal, but criminal has escaped chasing of the police by vehicle, like this for the public, still there is potential danger.Captured before criminal successfully escapes, become the most important thing of maintain social stability.Online at huge road traffic, the participant of traffic has several ten thousand even hundreds of thousands, and comprising walking, by bike, ride in a bus, to go by taxi or oneself is driven, the situation on road is fast changing.The concept of intelligent traffic monitoring system is arisen at the historic moment, and it makes the transport information on road and traffic-relevant information try one's best complete and real-time; Message exchange between traffic participant, traffic administration person, the vehicles, road management facility is real-time and efficient.Intelligent traffic monitoring system is exactly pass the image scene in monitor area back command centre by supervisory system, managerial personnel are made directly to grasp the traffics such as vehicle queue, blocking, signal lamp, timely adjustment signal timing dial or relieved traffic congestion by other means, change the distribution of traffic flow, to reach the object alleviating traffic jam.Even if there is science and technology of today to provide support, the suspected vehicles of criminal be found also to be a great problem in the so huge system of quantity of information.
And image recognition and this computer vision field of coupling have been developed fully, moving vehicles detection and tracking technology, as the important component part of safety assistant driving, intelligent transportation system research field has had tremendous development.Application number is the Chinese patent of 201210315966.5, patent name is: the vehicle checking method that a kind of Real-time Feedback upgrades, " comprise off-line learning process, in real time testing process and on-line study process; the 1st ~ K frame picture of the off-line strong classifier first utilizing off-line learning process to obtain to real-time testing process is classified, obtain detecting target; On-line study process detects target intercepted samples according to obtaining, and utilizes online strong classifier to carry out vehicle detection, obtains detecting target; On-line study process constantly upgrades online strong classifier.Application number is the Chinese patent of 201310014669.1, patent name is: the vehicle checking method in intelligent traffic monitoring system and device, " the present invention proposes the vehicle checking method in a kind of intelligent traffic monitoring system; it comprises: step S101, by charge coupled cell CCD camera collection road traffic sequence of frames of video; The sequence of frames of video data of collection are carried out Image semantic classification, are obtained the digital video sequences that computing machine can identify by step S102; The digital video sequences of step S103, input step S102 gained, utilizes the detection of the mixed Gaussian background modeling algorithm realization moving target improved; Step S104, moving target prospect step S103 being detected to gained carries out shadow Detection; Step S105, carries out shadow removal to moving target prospect, realizes the correct identification of moving target.But the method for moving vehicles detection and tracking is applied to locking, follows the trail of and catch and crime also belongs to blank.
Summary of the invention
In order to overcome the shortcoming of prior art existence with not enough, the invention provides a kind of method of recognition and tracking criminal vehicle.
The present invention adopts following technical scheme:
A method for recognition and tracking criminal vehicle, comprises the steps:
S1 obtains the image information of criminal's vehicle, determines the target vehicle wanting recognition and tracking;
S2 extracts the Invariance feature of target vehicle, and described Invariance feature comprises SURF feature;
Invariance feature is quantized morphogenesis characters describer by S3, and the foundation as match cognization criminal vehicle is transferred to CCC, is sent to each crime detection site by CCC;
The potential region of S4 estimating target vehicle position, the vehicle taken by monitoring camera and profiler are matched by the crime detection site in potential region, and the match is successful then reduces scope of reconnaissance;
Invariance feature is sent to the front cruiser in the scope of reconnaissance after reducing by S5 CCC,
The front cruiser that S6 receives target vehicle feature utilizes real-time monitor video in car, and carry out real-time Invariance feature coupling to front vehicles, when target vehicle Invariance feature and front vehicles, the match is successful, then implement to follow the tracks of.
Described S2 extracts the SURF feature of target vehicle, concrete employing surf recognizer.
The potential region of described estimating target vehicle position, is specially:
What S4.1 calculated target vehicle flees from distance d=vt, and wherein t is that target vehicle flees from the time from being found, and v is the speed of a motor vehicle of target vehicle;
The potential region that circle that distance is radius R 1 is exactly target vehicle position, for center of circle O1, is fled from the position that S4.2 is found with target vehicle.
Describedly reduce scope of reconnaissance, be specially: to detect that the position of the monitoring camera of target vehicle is for center of circle O2, with flee from circle that the difference of the distance of distance R1 and O1 to O2 is radius R 2 be exactly reduce after scope of reconnaissance.
Beneficial effect of the present invention:
(1) The present invention gives a kind of science, utilize the identification of existing traffic surveillance and control system resource and the method for track of offender's vehicle, chase criminal to the police and provide technical support, the success ratio of solving a case can be improved to a certain extent and maintain public order;
(2) the invention provides criminal's target vehicle information interaction framework of " feature information extraction-center-control transmission-branch acquisition of information ", make this using standard sequencing, have suitable enforceability;
(3) the present invention makes existing traffic surveillance and control system resource combine with identification art with ripe images match effectively, for solving the possibility that social concern better provides, and not needing brand-new technology hardware supported, utilizing existing resource;
(4) method of the extraction target Invariance feature information in the present invention is not limited only to extract surf feature, as long as recognizer can be made to identify target exactly namely can be used as in the present invention the method extracting feature, therefore the present invention has good adaptability.
Accompanying drawing explanation
Fig. 1 is workflow diagram of the present invention;
Fig. 2 is potential region of the present invention and the schematic diagram reducing rear scope of reconnaissance.
Embodiment
Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited thereto.
Embodiment
As shown in Figure 1, a kind of method of recognition and tracking criminal vehicle, comprises the steps:
S1 provides according to alarm people or the monitoring video of guilty place obtains the image information of criminal vehicle, determines the target vehicle wanting recognition and tracking;
S2 extracts the Invariance feature of target vehicle, and the present embodiment adopts SURF recognizer to extract the SURF feature of target vehicle;
Described surf recognizer, is specially:
The extraction of S2.1 unique point: utilize hessian matrix computations eigenwert α,
Hessian matrix is as follows:
Wherein L
xx(x, σ) is image g (σ) second derivative in x direction after gaussian filtering.
The eigenwert of S2.2 then hessian matrix is: det (H)=D
xxd
yy-(wD
xy)
2, get w=0.9. here
The SURF characteristic quantification morphogenesis characters describer that S3 will extract from target vehicle, as the foundation of match cognization criminal vehicle, and is transferred to CCC, is sent to each crime detection site by CCC;
The potential region of S4 estimating target vehicle position, the vehicle taken by monitoring camera and profiler are matched by the crime detection site in potential region, and the match is successful then reduces scope of reconnaissance;
Estimate the potential region of criminal's car two position, be specially:
What calculate target vehicle flees from distance d=vt, and wherein t is that target vehicle flees from the time from being found, and v is the speed of a motor vehicle of target vehicle;
The position be found with target vehicle is center of circle O1, flee from distance for the circle of radius R 1 be exactly the potential region of target vehicle position, as shown in phantom in Figure 2.
Described the match is successful then reduces scope of reconnaissance, is specially:
To detect that the position of the monitoring camera of target vehicle is for center of circle O2, with the difference of the distance of R1 and O1 to O2 for radius R 2 is justified, be the scouting region after reducing.As illustrated in solid line in figure 2.
SURF feature is sent to the front cruiser in the scouting region after reducing by S5 CCC;
The front cruiser that S6 receives target vehicle feature utilizes real-time monitor video in car, and monitor front vehicles, and carry out SURF characteristic matching in real time, when target vehicle SURF feature and front vehicles, the match is successful, then implement to follow the tracks of.
Described surf characteristic matching, is specially:
S6.1 point-of-interest extracts: calculate hessian determinant of a matrix, when the value of determinant is greater than threshold value, this point is extreme point.
S6.2 sets up metric space: the size changing the scale factor σ of gaussian filtering, does the gaussian filtering under different scale, set up the metric space of image to image.
It, to the point-of-interest obtained in (S6.1), compares with 18 points under eight points adjacent under current scale and adjacent yardstick, if this point is maximal value or minimum value, is then defined as unique point by S6.3.
S6.4 determines direction: we calculate the little wave response of Haar in x and y direction in the circle shaped neighborhood region of point of interest radius 6s, and s is the yardstick at point of interest place.Sampling step length does not rely on yardstick yet, elects s as.In order to other be consistent, small echo size does not rely on yardstick yet, length of side location 4s.After small echo RESPONSE CALCULATION is good, with the Gaussian filter weighting of σ=2s being centrally located at point of interest, the point of response in two-dimensional space represents, horizontal ordinate represents horizontal respone, and ordinate represents vertical response.Principal direction is by calculated response in the moving window that size is π/3 and estimate.Horizontal and vertical response in window is added, two and export a direction vector, and the most long vector in all windows defines the direction of point of interest.
S6.5 set up based on the little wave response of haar and descriptor: build a square region, center is at point of interest, and the direction that orientation is selected in upper joint, the size of window is 20s.This region by the neat subregion being divided into 4x4, wherein contains important spatial information again.To every sub regions, we calculate the little wave response of the equidistant sampled point of 5x5.Then, little wave response dx and dy of every sub regions is added, and constitutes first group of entry of proper vector.In order to introduce the polarity information of Strength Changes, we have extracted response absolute value simultaneously | dx| and | dy| and.Strength structure description vectors v=that thus, every sub regions has had 4 to tie up (Σ dx, Σ dy, Σ | dx|, Σ | dy|).Continuation like this calculates all 4x4 subregions, and result is exactly the description vectors of 64 dimensions, namely can be used as the feature of coupling.
Above-described embodiment is the present invention's preferably embodiment; but embodiments of the present invention are not limited by the examples; change, the modification done under other any does not deviate from Spirit Essence of the present invention and principle, substitute, combine, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.
Claims (4)
1. a method for recognition and tracking criminal vehicle, is characterized in that, comprises the steps:
S1 obtains the image information of criminal's vehicle, determines the target vehicle wanting recognition and tracking;
S2 extracts the Invariance feature of target vehicle, and described Invariance feature comprises SURF feature;
Invariance feature is quantized morphogenesis characters describer by S3, and the foundation as match cognization criminal vehicle is transferred to CCC, is sent to each crime detection site by CCC;
S4 estimates the potential region of criminal's vehicle position, and the vehicle taken by monitoring camera and profiler are matched by the crime detection site in potential region, and the match is successful then reduces scope of reconnaissance;
Invariance feature is sent to the front cruiser in the scope of reconnaissance after reducing by S5 CCC,
The front cruiser that S6 receives criminal's vehicle characteristics utilizes real-time monitor video in car, and carry out real-time Invariance feature coupling to front vehicles, when target vehicle Invariance feature and front vehicles, the match is successful, then implement to follow the tracks of.
2. the method for a kind of recognition and tracking criminal vehicle according to claim 1, is characterized in that, S2 extracts the SURF feature of target vehicle, concrete employing surf recognizer.
3. the method for a kind of recognition and tracking criminal vehicle according to claim 1, is characterized in that, the potential region of described estimation criminal's vehicle position, is specially:
What S4.1 calculated criminal flees from distance d=vt, and wherein t is that criminal flees from the time from being found, and v is the speed of a motor vehicle of criminal's vehicle;
The potential region that circle that distance is radius R 1 is exactly criminal's vehicle position, for center of circle O1, is fled from the position that S4.2 is found with criminal.
4. the method for a kind of recognition and tracking criminal vehicle according to claim 1, it is characterized in that, describedly reduce scope of reconnaissance, be specially: to detect that the position of the monitoring camera of target vehicle is for center of circle O2, the circle being radius R 2 with the radius R 1 in the potential region of criminal's vehicle and the difference of the distance of center of circle O1 to the O2 in potential region is exactly the scope of reconnaissance after reducing, described O1 is the position that criminal is found, and R1 size equals to flee from distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310714749.8A CN103714697B (en) | 2013-12-20 | 2013-12-20 | A kind of method of recognition and tracking criminal vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310714749.8A CN103714697B (en) | 2013-12-20 | 2013-12-20 | A kind of method of recognition and tracking criminal vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103714697A CN103714697A (en) | 2014-04-09 |
| CN103714697B true CN103714697B (en) | 2015-10-28 |
Family
ID=50407625
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310714749.8A Active CN103714697B (en) | 2013-12-20 | 2013-12-20 | A kind of method of recognition and tracking criminal vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103714697B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104574993B (en) * | 2014-12-30 | 2016-07-13 | 北京数字智通科技有限公司 | A kind of method of road monitoring and device |
| CN106157630B (en) * | 2016-08-01 | 2019-08-27 | 百度在线网络技术（北京）有限公司 | Monitoring method, vehicle and the system of suspicion of crime target |
| CN107545739A (en) * | 2017-09-06 | 2018-01-05 | 公安部道路交通安全研究中心 | A kind of method for controlling traffic signal lights and equipment |
| JP7047374B2 (en) * | 2017-12-25 | 2022-04-05 | トヨタ自動車株式会社 | Information gathering system |
| CN109325965A (en) * | 2018-08-22 | 2019-02-12 | 浙江大华技术股份有限公司 | A kind of target object tracking and device |
| CN109819207B (en) * | 2018-12-25 | 2020-07-21 | 深圳市天彦通信股份有限公司 | Target searching method and related equipment |
| CN109977909B (en) * | 2019-04-04 | 2021-04-20 | 山东财经大学 | Finger vein identification method and system based on minutia area matching |
| CN111651690A (en) * | 2020-05-29 | 2020-09-11 | 深圳市天一智联科技有限公司 | Case-related information searching method and device and computer equipment |
| CN112651992B (en) * | 2020-06-29 | 2024-04-05 | 浙江宇视科技有限公司 | Track tracking method and system |
| CN112507894A (en) * | 2020-12-14 | 2021-03-16 | 天时地理(深圳)智能科技有限公司 | Vehicle identification tracking method and system |
| US11699342B2 (en) | 2021-10-25 | 2023-07-11 | International Business Machines Corporation | Reducing vehicle occupant activity |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101656913A (en) * | 2009-09-23 | 2010-02-24 | 中兴通讯股份有限公司 | Method and system for monitoring analysis based on mobile network |
| CN201838185U (en) * | 2010-08-25 | 2011-05-18 | 浙江师范大学 | Vehicle detecting, tracking and recognizing system |
| CN102637257A (en) * | 2012-03-22 | 2012-08-15 | 北京尚易德科技有限公司 | Video-based detection and recognition system and method of vehicles |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011129800A1 (en) * | 2010-04-15 | 2011-10-20 | Milan Zlojutro | Vehicle monitoring & identification system |
-
2013
- 2013-12-20 CN CN201310714749.8A patent/CN103714697B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101656913A (en) * | 2009-09-23 | 2010-02-24 | 中兴通讯股份有限公司 | Method and system for monitoring analysis based on mobile network |
| CN201838185U (en) * | 2010-08-25 | 2011-05-18 | 浙江师范大学 | Vehicle detecting, tracking and recognizing system |
| CN102637257A (en) * | 2012-03-22 | 2012-08-15 | 北京尚易德科技有限公司 | Video-based detection and recognition system and method of vehicles |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103714697A (en) | 2014-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103714697B (en) | A kind of method of recognition and tracking criminal vehicle | |
| CN106874863B (en) | Vehicle illegal parking and reverse running detection method based on deep convolutional neural network | |
| CN202077142U (en) | Vehicle-mounted intelligent video detecting and analyzing system | |
| CN105702048B (en) | Highway front truck illegal road occupation identifying system based on automobile data recorder and method | |
| CN106384513B (en) | A kind of fake-licensed car capture system and method based on intelligent transportation | |
| CN106203274A (en) | Pedestrian's real-time detecting system and method in a kind of video monitoring | |
| CN106204640A (en) | A kind of moving object detection system and method | |
| CN103810717B (en) | A kind of human body behavioral value method and device | |
| CN111860318A (en) | Construction site pedestrian loitering detection method, device, equipment and storage medium | |
| CN105574506A (en) | Intelligent face tracking system and method based on depth learning and large-scale clustering | |
| CN106485245A (en) | A kind of round-the-clock object real-time tracking method based on visible ray and infrared image | |
| CN102254394A (en) | Antitheft monitoring method for poles and towers in power transmission line based on video difference analysis | |
| CN105321352A (en) | A motor vehicle license plate blocking violation detection and evidence obtaining method | |
| CN103000029B (en) | Automobile video frequency recognition methods and application thereof | |
| Xu et al. | Efficient CityCam-to-edge cooperative learning for vehicle counting in ITS | |
| CN112084928A (en) | Road traffic accident detection method based on visual attention mechanism and ConvLSTM network | |
| CN103426179A (en) | Target tracking method and system based on mean shift multi-feature fusion | |
| CN105657372A (en) | Method and system for realizing intelligent detection and early warning of on-duty guard posts by videos | |
| CN113743260A (en) | Pedestrian tracking method under dense pedestrian flow condition of subway platform | |
| CN101635026A (en) | Method for detecting derelict without tracking process | |
| Luo et al. | Vehicle flow detection in real-time airborne traffic surveillance system | |
| CN106960193A (en) | A kind of lane detection apparatus and method | |
| CN105206060A (en) | Vehicle type recognition device and method based on SIFT characteristics | |
| CN103426178A (en) | Target tracking method and system based on mean shift in complex scene | |
| CN101499214A (en) | Automatic traffic parameter extraction method based on image information entropy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |