CN104715264A - Method and system for recognizing video images of motion states of vehicles in expressway tunnel - Google Patents

Method and system for recognizing video images of motion states of vehicles in expressway tunnel Download PDF

Info

Publication number
CN104715264A
CN104715264A CN201510166895.0A CN201510166895A CN104715264A CN 104715264 A CN104715264 A CN 104715264A CN 201510166895 A CN201510166895 A CN 201510166895A CN 104715264 A CN104715264 A CN 104715264A
Authority
CN
China
Prior art keywords
vehicle
tunnel
geometric properties
video
video 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.)
Pending
Application number
CN201510166895.0A
Other languages
Chinese (zh)
Inventor
陈小佳
杨科
崔太雷
苏明义
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong Traffic Splendid Jiangxi South Guangdong Province Highway Control Center
Wuhan University of Technology WUT
Original Assignee
Guangdong Traffic Splendid Jiangxi South Guangdong Province Highway Control Center
Wuhan University of Technology WUT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangdong Traffic Splendid Jiangxi South Guangdong Province Highway Control Center, Wuhan University of Technology WUT filed Critical Guangdong Traffic Splendid Jiangxi South Guangdong Province Highway Control Center
Priority to CN201510166895.0A priority Critical patent/CN104715264A/en
Publication of CN104715264A publication Critical patent/CN104715264A/en
Pending legal-status Critical Current

Links

Landscapes

  • Image Analysis (AREA)

Abstract

The invention discloses a method and system for recognizing video images of motion states of vehicles in an expressway tunnel. The method comprises the following steps: carrying out partition on a video image of a vehicle moving object based on background modeling, and respectively carrying out extraction on multiple features of the vehicle image, wherein the features include a texture feature, a geometric feature and an edge feature; constructing a neural network topology architecture as a base classifier by using the extracted multiple features; carrying out integration on the base classifier by using an Adaboost method so as to form a strong classifier; carrying out vehicle image recognition through the strong classifier; and when the existence of a parking state is recognized, sending a signal to a video monitoring center, automatically switching the monitoring center to monitor the image, starting an intra-tunnel parking state alarm, and releasing a notice to the outside of a tunnel. According to the invention, the hazardous state of vehicle parking in the tunnel can be intelligently recognized by using images transmitted from a video surveillance camera back to the monitoring center, and an alarm is made in time, thereby improving the automation level of the management of the expressway tunnel.

Description

State of motion of vehicle video image identification method and system in freeway tunnel
Technical field
The present invention relates to image procossing, particularly relate to state of motion of vehicle video image identification method and system in a kind of freeway tunnel.
Background technology
Video image vehicle identification is the low advantage with being convenient to implement due to its cost, becomes the focus of freeway management application gradually.Current vehicle identification system can meet the basic function demand that the intelligent transportation systems such as wagon flow statistics, overspeed detection are paid close attention to substantially.But at some particular places of highway (as tunnel), by vehicle identification accuracy rate and the rate of false alarm of video image and utilizing the application of this technology still to there is more deficiency, in theory and technically also there is a lot of problem and do not solve.
To stop and the major traffic accidents that cause had many reports at home and abroad because vehicle trouble, accident or block up causes in freeway tunnel.If can be monitored automatically by head end video in the very first time, obtain situation in tunnel exactly, and issue tunnel internal information in time, informing the vehicle being about to enter tunnel, having important practical significance to avoiding rear-end collision in tunnel.Carry out omnidistance video monitoring in China's freeway tunnel, on monitoring hardware, function is comparatively perfect, but the monitor screen that Surveillance center is limited, rely on and be manually difficult to observe the situation in tunnel comprehensively, make the benefit of monitoring hardware be difficult to play.
The factors such as the illumination of freeway tunnel background condition further aspect is not enough and illumination variation is strong cause vehicle identification difficulty; On the other hand the various Different factor such as vehicle shape, size, color and camera angle also makes to develop general vehicle identification system and remains a more difficult problem.Study vehicle identification from the angle of vehicle tracking and classification problem studies more approach both at home and abroad at the past period, this kind of method comparison is ripe, but can not distinguish other objects preferably; In addition also have the research of vehicle type recognition aspect, these class methods lay particular emphasis on vehicle type recognition, and under complex background, effect is undesirable; And under complex background, mainly concentrating on vehicle at night identification, these methods mainly utilize the identification of automobile tail light to detect vehicle.There is the principal component method of document utilization image to carry out feature extraction, then adopt support vector machine (SVM) sorter to carry out vehicle detection.The feature extraction of nearly several years problem of image recognition deflection object, characteristics of image comprises visual signature, statistical nature, transform domain feature etc.As adopted the video encoder server algorithm based on Haar_like rectangular characteristic, first the method extracts Haar_like rectangular characteristic, then utilizes the Adaboost method of improvement to carry out integrated.For vehicle identification problem, adopt the feature of single class, existing characteristics descriptive power is not enough, the problem that recognition performance is undesirable.Therefore need to research and develop the vehicle recongnition technique with pervasive ability.
Summary of the invention
The technical problem to be solved in the present invention is the feature for adopting single class in prior art, and existing characteristics descriptive power is not enough, and the defect that recognition performance is undesirable, provides state of motion of vehicle video image identification method and system in a kind of freeway tunnel.
The technical solution adopted for the present invention to solve the technical problems is:
A kind of state of motion of vehicle video image identification method in freeway tunnel is provided, comprises the following steps:
Background modeling basis is split vehicle movement target video image, respectively multiple features of vehicle image is extracted, comprise textural characteristics, geometric properties and edge feature;
By the multiple feature construction neural network topology structure extracted, as base sorter;
Utilize Adaboost method to carry out integrated to base sorter, form strong classifier;
Vehicle image identification is carried out by strong classifier.
In method of the present invention, described textural characteristics represents with 4 kinds of characteristic parameters of gray scale symbiosis square, is second moment, moment of inertia, unfavourable balance square and entropy respectively;
Described geometric properties is with 7 variablees composition of Hu not bending moment, and it is constructed by centre distance, has the geometric properties of rotation, translation and scale invariance;
Described edge feature describes with the low-and high-frequency energy after db2 wavelet transformation.
In method of the present invention, utilize textural characteristics, geometric properties and edge feature to build neural network topology structure as base sorter, this neural network topology structure comprises 13 input nodes, 13 hidden nodes and an output node.
Present invention also offers state of motion of vehicle video image identification system in a kind of freeway tunnel, comprising:
Characteristic extracting module, for splitting vehicle movement target video image on background modeling basis, extracting multiple features of vehicle image respectively, comprising textural characteristics, geometric properties and edge feature;
Base classifier modules, for the multiple feature construction neural network topology structure that will extract, as base sorter;
Strong classifier module, for utilizing Adaboost method to carry out integrated to base sorter, forms strong classifier;
Picture recognition module, for carrying out vehicle image identification by strong classifier.
In system of the present invention, described characteristic extracting module represents with 4 kinds of characteristic parameters of gray scale symbiosis square specifically for the described textural characteristics extracted, and is second moment, moment of inertia, unfavourable balance square and entropy respectively;
The described geometric properties extracted is with 7 variablees composition of Hu not bending moment, and it is constructed by centre distance, has the geometric properties of rotation, translation and scale invariance;
The described edge feature extracted describes with the low-and high-frequency energy after db2 wavelet transformation.
In system of the present invention, described base classifier modules specifically utilizes textural characteristics, geometric properties and edge feature to build neural network topology structure as base sorter, and this neural network topology structure comprises 13 input nodes, 13 hidden nodes and an output node.
Present invention also offers vehicle in a kind of freeway tunnel, because of the abnormal response processing method causing stopping, to comprise the following steps:
Obtain vision signal;
Video image identification method according to claim 1 identifies the vision signal obtained;
When recognize there is dead ship condition time, send signal to video monitoring center, automatic switchover Surveillance center monitoring picture, and to start in tunnel dead ship condition and report to the police, make an announcement outside tunnel.
Present invention also offers vehicle in a kind of freeway tunnel, because of the abnormal response treating apparatus causing stopping, to comprise:
Video acquiring module, for obtaining vision signal;
Video image identification system, identifies the vision signal obtained for video image identification method according to claim 1;
Alarm module, for when recognize there is dead ship condition time, send signal to video monitoring center, automatic switchover Surveillance center monitoring picture, and to start in tunnel dead ship condition and report to the police, make an announcement outside tunnel.
The beneficial effect that the present invention produces is: multiclass feature combines by the present invention, strengthen the descriptive power of model of cognition, comprise vehicle image textural characteristics, vehicle geometric properties and vehicle edge characteristics of image, then utilize that the Adaboost of BP neural network is integrated sets up vehicle identification model.The present invention identifies the state of motion of vehicle in freeway tunnel emphatically, and automatically makes a response to wherein stopping because of abnormal conditions and provide warning, can greatly improve high-speed secure management level.
Accompanying drawing explanation
Below in conjunction with drawings and Examples, the invention will be further described, in accompanying drawing:
Fig. 1 is the process flow diagram of state of motion of vehicle video image identification method in embodiment of the present invention freeway tunnel;
Fig. 2 be in the embodiment of the present invention freeway tunnel vehicle because of the abnormal response processing method process flow diagram causing stopping;
Fig. 3 is typical dolly video scene in embodiment of the present invention tunnel;
Fig. 4 is high capacity waggon video scene in embodiment of the present invention tunnel;
Fig. 5 is state of motion of vehicle video image identification system architecture schematic diagram in embodiment of the present invention freeway tunnel;
Fig. 6 be in the embodiment of the present invention freeway tunnel vehicle because of the abnormal response treating apparatus structural representation causing stopping.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
In freeway tunnel of the present invention, state of motion of vehicle video image identification method, is through the steps such as video image acquisition, image procossing and image recognition, to judge state of motion of vehicle, identifies dead ship condition, and then issues warning instruction.
As shown in Figure 1, comprise the following steps:
S11, on background modeling basis, vehicle movement target video image to be split, respectively multiple features of vehicle image are extracted, comprise textural characteristics, geometric properties and edge feature;
S12, the multiple feature construction neural network topology structure that will extract, as base sorter;
S13, utilize Adaboost method to carry out integrated to base sorter, form strong classifier;
S14, carry out vehicle image identification by strong classifier.
In step S11, by gradation of image co-occurrence matrix, the Hu not computation model such as bending moment statistical nature and Gray-level co-occurrence, extract the multiple vehicle characteristics such as vehicle image textural characteristics, vehicle geometric properties and vehicle edge characteristics of image respectively.Merge vehicle characteristics from multiple angles, improve vehicle identification accuracy rate and decrease rate of false alarm.The concrete computation model such as vehicle image textural characteristics, vehicle geometric properties and vehicle edge characteristics of image is as follows:
Vehicle image textural characteristics is characterized by the second moment of gradation of image co-occurrence matrix, moment of inertia, unfavourable balance square and entropy, respectively:
Second moment: f 1 = Σ i = 0 L - 1 Σ j = 0 L - 1 P 2 ( i , j )
Moment of inertia: f 2 = Σ i = 0 L - 1 Σ j = 0 L - 1 ( i - j ) 2 P ( i , j )
Unfavourable balance square: f 3 = Σ i = 0 L - 1 Σ j = 0 L - 1 P ( i , j ) 1 + ( i - j ) 2
Entropy: f 4 = Σ i = 0 L - 1 Σ j = 0 L - 1 P ( i , j ) log P ( i , j )
In formula, P (i, j) is gradation of image co-occurrence matrix.
Vehicle geometric properties is by having seven variablees formations of rotation, translation and scale invariance in image Hu not bending moment statistical nature, second order and third central moment is used to constitute seven to all insensitive Character eigenvector of image rotation, translation, convergent-divergent, respectively:
M 1＝φ 20+φ 02
M 2 = ( φ 20 - φ 02 ) 2 + 4 φ 11 2
M 3＝(φ 30-3φ 12) 2+(φ 21-3φ 03) 2
M 4＝(φ 30+φ 12) 2+(φ 21+φ 03) 2
M 5＝(φ 30-3φ 12)(φ 30+φ 12)[(φ 30+φ 12) 2-3(φ 21+φ 03) 2]
+(3φ 21-φ 03)(φ 21+φ 03)[3(φ 30+φ 12) 2-(φ 21+φ 03) 2]
M 6＝(φ 20-φ 02)[(φ 30+φ 12) 2-(φ 21+φ 03) 2]+4φ 11(φ 30+φ 12)(φ 21+φ 03)
M 7＝(3φ 21-φ 03)(φ 30+φ 12)[(φ 30+φ 12) 2-3(φ 21+φ 03) 2]
+(3φ 12-φ 30)(φ 21+φ 03)[3(φ 30+φ 12) 2-(φ 21+φ 03) 2]
In formula for normalization center, (p+q) rank square.These 7 proper vectors are the quantic of normalization center, (p+q) rank square, as φ 20, be p=2, q=0, φ 11in be p=1, q=1, the like.
Vehicle edge characteristics of image is characterized by Gray-level co-occurrence.In video image, vehicle region is compared with corresponding background, and the edge in this region and detailed information are increased, and namely vehicle edge is corresponding with the high-frequency information in image.Consider the relativity of high-frequency energy and low frequency energy, adopt the low-and high-frequency energy after db2 wavelet transformation to describe vehicle image feature.Be expressed as:
L＝EL_ROI/EL
H＝EH_ROI/EH
In formula, EL, EH, EL_ROI and EH_ROI are respectively background area low frequency gross energy, background area high frequency gross energy, region-of-interest low frequency gross energy and region-of-interest high frequency gross energy.
In step S13, iterated by Weak Classifier and obtain several classification function sequences, and give a weight to each classification function, according to classifying quality quality, distribute the size of corresponding weight.After successive ignition, final strong classification function is obtained by the weighting of weak typing function.Using BP network as Weak Classifier, repetition training BP neural network forecast ability, obtains the strong classifier of multiple BP network Weak Classifier composition by Adaboost algorithm.
In step S14, the strong classifier obtained by step S13, carries out Classification and Identification to the situation that there is vehicle in video image.
In preferred embodiment of the present invention, build neural network topology structure as base sorter using above-mentioned vehicle characteristics, utilize Adaboost method to carry out integrated, form strong classifier (10) for vehicle image identification.Specifically comprise:
With above-mentioned 4 textural characteristics, 7 Hu moment preserving square statistical natures and 2 Gray-level co-occurrence, build the neural network topology structure of 13 input nodes, 13 hidden nodes and an output node as base sorter;
Utilize Adaboost method to carry out integrated, form strong classifier and be used for vehicle image identification.
As shown in Figure 2, present invention also offers a kind of solution in freeway tunnel and identify that vehicle is because of the abnormal system responses disposal route causing stopping, and comprises step in time:
S21, acquisition vision signal;
S22, according to the video image identification method of above-described embodiment, the vision signal obtained to be identified;
S23, when recognize there is dead ship condition time, signal is sent to video monitoring center, automatic switchover Surveillance center monitoring picture, and start dead ship condition warning in tunnel, make an announcement outside tunnel, this PASS VIDEO image record switch can also be started simultaneously, and preserve respective channel video image, consult in order to follow-up.
A specific embodiment of the present invention is as follows:
Fig. 3 and Fig. 4 is typical dolly and high capacity waggon video scene in certain tunnel, adopts the technology of the present invention, and simulation is stopped, the accuracy of checking auto model.
Following table 1 is the contrast situation of the present invention and other two kinds of models, illustrates that the present invention's just inspection rate, false drop rate, loss and recognition accuracy are all more excellent.
The contrast situation of table 1 the present invention and other two kinds of models
Present invention also offers state of motion of vehicle video image identification system in a kind of freeway tunnel, can solve the many and Surveillance center's display screen quantity of video frequency pick-up head number of active lanes in freeway tunnel few between contradiction, can the parking that causes of the multiple situation of Intelligent Recognition within a short period of time tunnel internal cause abnormal, and automatically switch to Surveillance center's screen, and corresponding startup alarm signal, avoid having an accident in tunnel, improve highway automatic management level.As shown in Figure 5, this image identification system comprises:
Characteristic extracting module, for splitting vehicle movement target video image on background modeling basis, extracting multiple features of vehicle image respectively, comprising textural characteristics, geometric properties and edge feature;
Base classifier modules, for the multiple feature construction neural network topology structure that will extract, as base sorter;
Strong classifier module, for utilizing Adaboost method to carry out integrated to base sorter, forms strong classifier;
Picture recognition module, for carrying out vehicle image identification by strong classifier.
Characteristic extracting module represents with 4 kinds of characteristic parameters of gray scale symbiosis square specifically for the described textural characteristics extracted, and is second moment, moment of inertia, unfavourable balance square and entropy respectively;
The described geometric properties extracted is with 7 variablees composition of Hu not bending moment, and it is constructed by centre distance, has the geometric properties of rotation, translation and scale invariance;
The described edge feature extracted describes with the low-and high-frequency energy after db2 wavelet transformation.
Base classifier modules specifically utilizes textural characteristics, geometric properties and edge feature to build neural network topology structure as base sorter, and this neural network topology structure comprises 13 input nodes, 13 hidden nodes and an output node.
Present invention also offers vehicle in a kind of freeway tunnel because of the abnormal response treating apparatus causing stopping, can realize not changing on existing video monitoring system hardware foundation, there is applicability widely.As shown in Figure 6, comprising:
Video acquiring module, for obtaining vision signal;
Video image identification system, identifies the vision signal obtained for video image identification method according to claim 1;
Alarm module, for when recognize there is dead ship condition time, send signal to video monitoring center, automatic switchover Surveillance center monitoring picture, and to start in tunnel dead ship condition and report to the police, make an announcement outside tunnel.
Should be understood that, for those of ordinary skills, can be improved according to the above description or convert, and all these improve and convert the protection domain that all should belong to claims of the present invention.

Claims (8)

1. a state of motion of vehicle video image identification method in freeway tunnel, is characterized in that, comprise the following steps:
Background modeling basis is split vehicle movement target video image, respectively multiple features of vehicle image is extracted, comprise textural characteristics, geometric properties and edge feature;
By the multiple feature construction neural network topology structure extracted, as base sorter;
Utilize Adaboost method to carry out integrated to base sorter, form strong classifier;
Vehicle image identification is carried out by strong classifier.
2. method according to claim 1, is characterized in that, described textural characteristics represents with 4 kinds of characteristic parameters of gray scale symbiosis square, is second moment, moment of inertia, unfavourable balance square and entropy respectively;
Described geometric properties is with 7 variablees composition of Hu not bending moment, and it is constructed by centre distance, has the geometric properties of rotation, translation and scale invariance;
Described edge feature describes with the low-and high-frequency energy after db2 wavelet transformation.
3. method according to claim 1, it is characterized in that, utilize textural characteristics, geometric properties and edge feature to build neural network topology structure as base sorter, this neural network topology structure comprises 13 input nodes, 13 hidden nodes and an output node.
4. a state of motion of vehicle video image identification system in freeway tunnel, is characterized in that, comprising:
Characteristic extracting module, for splitting vehicle movement target video image on background modeling basis, extracting multiple features of vehicle image respectively, comprising textural characteristics, geometric properties and edge feature;
Base classifier modules, for the multiple feature construction neural network topology structure that will extract, as base sorter;
Strong classifier module, for utilizing Adaboost method to carry out integrated to base sorter, forms strong classifier;
Picture recognition module, for carrying out vehicle image identification by strong classifier.
5. system according to claim 4, is characterized in that, described characteristic extracting module represents with 4 kinds of characteristic parameters of gray scale symbiosis square specifically for the described textural characteristics extracted, and is second moment, moment of inertia, unfavourable balance square and entropy respectively;
The described geometric properties extracted is with 7 variablees composition of Hu not bending moment, and it is constructed by centre distance, has the geometric properties of rotation, translation and scale invariance;
The described edge feature extracted describes with the low-and high-frequency energy after db2 wavelet transformation.
6. system according to claim 4, it is characterized in that, described base classifier modules specifically utilizes textural characteristics, geometric properties and edge feature to build neural network topology structure as base sorter, and this neural network topology structure comprises 13 input nodes, 13 hidden nodes and an output node.
7. in freeway tunnel, vehicle, because of the abnormal response processing method causing stopping, is characterized in that, comprises the following steps:
Obtain vision signal;
Video image identification method according to claim 1 identifies the vision signal obtained;
When recognize there is dead ship condition time, send signal to video monitoring center, automatic switchover Surveillance center monitoring picture, and to start in tunnel dead ship condition and report to the police, make an announcement outside tunnel.
8. in freeway tunnel, vehicle, because of the abnormal response treating apparatus causing stopping, is characterized in that, comprising:
Video acquiring module, for obtaining vision signal;
Video image identification system, identifies the vision signal obtained for video image identification method according to claim 1;
Alarm module, for when recognize there is dead ship condition time, send signal to video monitoring center, automatic switchover Surveillance center monitoring picture, and to start in tunnel dead ship condition and report to the police, make an announcement outside tunnel.
CN201510166895.0A 2015-04-10 2015-04-10 Method and system for recognizing video images of motion states of vehicles in expressway tunnel Pending CN104715264A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510166895.0A CN104715264A (en) 2015-04-10 2015-04-10 Method and system for recognizing video images of motion states of vehicles in expressway tunnel

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510166895.0A CN104715264A (en) 2015-04-10 2015-04-10 Method and system for recognizing video images of motion states of vehicles in expressway tunnel

Publications (1)

Publication Number Publication Date
CN104715264A true CN104715264A (en) 2015-06-17

Family

ID=53414574

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510166895.0A Pending CN104715264A (en) 2015-04-10 2015-04-10 Method and system for recognizing video images of motion states of vehicles in expressway tunnel

Country Status (1)

Country Link
CN (1) CN104715264A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106815558A (en) * 2016-12-21 2017-06-09 上海智驾电子科技有限公司 Car based on image recognition enters tunnel headlight and shifts to an earlier date open method automatically
CN107464428A (en) * 2017-09-22 2017-12-12 长安大学 A kind of tunnel traffic anomalous identification warning device
CN108229341A (en) * 2017-12-15 2018-06-29 北京市商汤科技开发有限公司 Sorting technique and device, electronic equipment, computer storage media, program
CN108922172A (en) * 2018-06-19 2018-11-30 上海理工大学 Congestion in road based on vehicle characteristics matrix sequence mutation analysis monitors system
CN110058263A (en) * 2019-04-16 2019-07-26 广州大学 A kind of object positioning method in vehicle travel process
CN110084112A (en) * 2019-03-20 2019-08-02 太原理工大学 A kind of traffic congestion judgment method based on image procossing
CN111429756A (en) * 2020-03-30 2020-07-17 淮阴工学院 Highway tunnel rear-end collision prevention early warning method based on video technology

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7054491B2 (en) * 2001-11-16 2006-05-30 Stmicroelectronics, Inc. Scalable architecture for corresponding multiple video streams at frame rate
KR100896949B1 (en) * 2006-08-30 2009-05-14 (주)로직아이텍 Image Monitoring System for Object Identification
CN104299008A (en) * 2014-09-23 2015-01-21 同济大学 Vehicle type classification method based on multi-feature fusion

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7054491B2 (en) * 2001-11-16 2006-05-30 Stmicroelectronics, Inc. Scalable architecture for corresponding multiple video streams at frame rate
KR100896949B1 (en) * 2006-08-30 2009-05-14 (주)로직아이텍 Image Monitoring System for Object Identification
CN104299008A (en) * 2014-09-23 2015-01-21 同济大学 Vehicle type classification method based on multi-feature fusion

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
刘艳丽: ""基于神经网络的运动目标检测与跟踪算法的研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *
孙棣华: "基于视频检测技术的隧道停车检测与识别算法", 《计算机测量与控制》 *
张和颖: "基于视频的车辆检测与跟踪", 《中国优秀硕士学位论文全文数据库》 *
王正玉: "交通监控视频中机动车辆检索的关键技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106815558A (en) * 2016-12-21 2017-06-09 上海智驾电子科技有限公司 Car based on image recognition enters tunnel headlight and shifts to an earlier date open method automatically
CN106815558B (en) * 2016-12-21 2020-06-30 上海智驾汽车科技有限公司 Automatic early starting method for tunnel headlamp of vehicle based on image recognition
CN107464428A (en) * 2017-09-22 2017-12-12 长安大学 A kind of tunnel traffic anomalous identification warning device
CN108229341A (en) * 2017-12-15 2018-06-29 北京市商汤科技开发有限公司 Sorting technique and device, electronic equipment, computer storage media, program
CN108229341B (en) * 2017-12-15 2021-08-06 北京市商汤科技开发有限公司 Classification method and device, electronic equipment and computer storage medium
CN108922172A (en) * 2018-06-19 2018-11-30 上海理工大学 Congestion in road based on vehicle characteristics matrix sequence mutation analysis monitors system
CN108922172B (en) * 2018-06-19 2021-03-05 上海理工大学 Road congestion monitoring system based on vehicle characteristic matrix sequence change analysis
CN110084112A (en) * 2019-03-20 2019-08-02 太原理工大学 A kind of traffic congestion judgment method based on image procossing
CN110084112B (en) * 2019-03-20 2022-09-20 太原理工大学 Traffic jam judging method based on image processing
CN110058263A (en) * 2019-04-16 2019-07-26 广州大学 A kind of object positioning method in vehicle travel process
CN111429756A (en) * 2020-03-30 2020-07-17 淮阴工学院 Highway tunnel rear-end collision prevention early warning method based on video technology
CN111429756B (en) * 2020-03-30 2022-02-11 淮阴工学院 Highway tunnel rear-end collision prevention early warning method based on video technology

Similar Documents

Publication Publication Date Title
CN104715264A (en) Method and system for recognizing video images of motion states of vehicles in expressway tunnel
Amato et al. Counting vehicles with deep learning in onboard UAV imagery
CN112633176B (en) Rail transit obstacle detection method based on deep learning
CN110717387B (en) Real-time vehicle detection method based on unmanned aerial vehicle platform
CN103902976A (en) Pedestrian detection method based on infrared image
CN103714697B (en) A kind of method of recognition and tracking criminal vehicle
CN106339657B (en) Crop straw burning monitoring method based on monitor video, device
CN109993138A (en) A kind of car plate detection and recognition methods and device
CN109508395B (en) Intelligent series-parallel plan method based on vehicle type retrieval
CN105404859A (en) Vehicle type recognition method based on pooling vehicle image original features
Karni et al. Development of autonomous downscaled model car using neural networks and machine learning
Chen et al. Vehicles detection on expressway via deep learning: Single shot multibox object detector
CN103927875A (en) Traffic overflowing state recognition method based on video
Amrutha et al. Real-time litter detection system for moving vehicles using YOLO
Rokonuzzaman et al. Automatic vehicle identification system using machine learning and robot operating system (ROS)
CN116129293A (en) Vehicle real-time detection method based on unmanned aerial vehicle
Dongtao et al. Traffic sign detection method of improved ssd based on deep learning
Soleimani et al. Convolutional neural networks for aerial vehicle detection and recognition
Ebren et al. Determining the occupancy of Vehicle Parking Areas by deep learning
CN105206060A (en) Vehicle type recognition device and method based on SIFT characteristics
Xia et al. Research on Traffic Accident Detection Based on Vehicle Perspective
Ullah et al. Moving vehicle detection and information extraction based on deep neural network
Wang et al. Abnormal event detection algorithm based on dual attention future frame prediction and gap fusion discrimination
Saripan et al. Tree-based vehicle classification system
Saadeldin et al. Real-time vehicle counting using custom YOLOv8n and DeepSORT for resource-limited edge devices

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20150617

RJ01 Rejection of invention patent application after publication