CN106778648B - Vehicle track tracking and license plate recognition method - Google Patents
Vehicle track tracking and license plate recognition method Download PDFInfo
- Publication number
- CN106778648B CN106778648B CN201611216113.0A CN201611216113A CN106778648B CN 106778648 B CN106778648 B CN 106778648B CN 201611216113 A CN201611216113 A CN 201611216113A CN 106778648 B CN106778648 B CN 106778648B
- Authority
- CN
- China
- Prior art keywords
- license plate
- result
- positioning
- module
- vehicle
- 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
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000012544 monitoring process Methods 0.000 claims abstract description 45
- 238000012545 processing Methods 0.000 claims abstract description 13
- 238000004891 communication Methods 0.000 claims abstract description 4
- 230000003287 optical effect Effects 0.000 claims abstract description 4
- 238000001514 detection method Methods 0.000 claims description 21
- 230000011218 segmentation Effects 0.000 claims description 18
- 238000012937 correction Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 6
- 230000003068 static effect Effects 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 241000233855 Orchidaceae Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/63—Scene text, e.g. street names
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/148—Segmentation of character regions
- G06V30/153—Segmentation of character regions using recognition of characters or words
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/625—License plates
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a vehicle track tracking and license plate recognition system, which comprises close-up cameras arranged on monitoring points, wherein the close-up cameras are connected with a downstream storage module arranged on each monitoring point through an optical transceiver and an encoder, the downstream storage module is in communication connection with a cloud computing platform, the cloud computing platform comprises a license plate analysis processing module special for analyzing and recognizing license plates, and the vehicle track tracking and license plate recognition system can quickly and accurately detect the license plates of running vehicles, so that the accurate identification and tracking of the vehicles are realized, and the safety performance is improved. In addition, the invention also discloses a recognition method of the recognition system, the recognition method firstly selects the video in the optimal time period as an analysis object, then actively analyzes the image of each frame, and finally compares the single character after dividing the character, thereby greatly improving the recognition accuracy and adapting to the requirement of active monitoring in the whole time period.
Description
Technical Field
The invention relates to a vehicle track tracking and license plate recognition system, and also relates to a recognition method of the vehicle track tracking and license plate recognition system.
Background
In the current urban internet of things management system, the requirement on security is higher and higher, particularly, the monitoring of vehicles is carried out, for example, after some flees to drive and escape, the vehicles cannot be tracked in time, the current security system can only be judged manually according to the colors and the styles of the vehicles, the automatic monitoring cannot be realized, the further accurate real-time tracking cannot be realized, and the running tracks of the vehicles cannot be tracked.
Disclosure of Invention
The first technical problem to be solved by the invention is as follows: the vehicle track tracking and license plate recognition system can quickly and accurately detect the license plate of a running vehicle, so that the accurate identification and tracking of the vehicle are realized, and the safety performance is improved.
The second technical problem to be solved by the invention is: the recognition method can monitor a running vehicle, segment the license plate after positioning the license plate of the vehicle, and accurately recognize the segmented license plate, and is rapid in recognition and high in accuracy.
In order to solve the first technical problem, the technical scheme of the invention is as follows: the utility model provides a vehicle trajectory tracking and license plate recognition system, is including setting up the close-up camera on each monitoring point, the close-up camera passes through optical transceiver and encoder and is connected with the low reaches storage module that sets up on every monitoring point, and this low reaches storage module is connected with the communication of cloud computing platform, including being exclusively used in the license plate analysis and processing module of analysis discernment license plate among the cloud computing platform, this license plate analysis and processing module includes
The vehicle detection tracking module is used for analyzing the video shot by the close-up camera, and determining the optimal time period as a video analysis object according to the definition of the license plate and the size of the license plate display area after tracking the position of the vehicle;
the license plate positioning module analyzes each frame image in the video analysis object in a mode of taking a license plate frame as a distinguishing characteristic and combining license plate ground color comparison, and determines the specific position of the license plate to obtain a positioning result image;
the license plate correction and fine positioning module corrects the positioning result picture and improves the sharpness and definition of the image, then accurately positions the positioning result of each frame of image, and removes irrelevant image parts again to obtain a fine positioning result picture;
the license plate segmentation module segments the fine positioning result picture of each frame of image into a plurality of pictures to be recognized, and each picture to be recognized only contains one character;
the character recognition module compares each picture to be recognized with the license plate characters stored in the database in advance one by one to obtain a primary recognition result;
the result decision module is used for obtaining the comprehensive reliability of the license plate according to the recording information of the license plate in the video process and the maximum reliability value of the primary recognition result obtained by combining each frame of picture, and obtaining a conclusion of outputting the recognition result or refusing the result to continuously track the license plate;
and the output management module outputs the identification result obtained by the result decision module.
As a preferred scheme, the license plate analysis processing module further comprises a learning module and a license plate tracking module; the license plate tracking module feeds back a positioning result picture obtained by the license plate positioning module to the video to determine the motion track of the license plate in the video when the vehicle runs; the learning module is connected with the vehicle detection and tracking module, the license plate positioning module, the license plate correction and fine positioning module, the license plate segmentation module, the character recognition module, the result decision module and the license plate tracking module, after the learning module analyzes and processes the license plates for multiple times, the learning module combines the motion track of the license plates in the video, the angle of the shot video, the picture quality and the vehicle running speed and the surrounding environment of the monitoring points to perform feedback guidance on the modules, so that the license plate positioning module preferentially identifies the position of the license plate track in each frame of picture, and guides other modules to perform adaptive optimization adjustment according to the shooting angle and the shooting environment of the monitoring points.
As a preferable scheme, the monitoring points include entrance and exit monitoring points of roads, entrance and exit monitoring points of public places, and entrance and exit monitoring points of districts and schools, wherein the number of close-up cameras on the monitoring points of the entrance and the exit of a road corresponds to the number of lanes of the road one by one, and panoramic cameras and auxiliary light sources are arranged on the monitoring points of the entrance and the exit of the road, wherein the close-up cameras arranged on the entrance and the exit monitoring points of the same road respectively shoot the head and the tail of a vehicle.
After the technical scheme is adopted, the invention has the effects that: the vehicle track tracking and license plate recognition system is an intelligent video processing system integrating a plurality of functions of image digital acquisition, storage and real-time automatic license plate recognition, and can be widely applied to monitoring and alarming of road passing vehicles. The system realizes modernization and intellectualization of public security management, provides a powerful intelligent monitoring system, and plays an important role in construction of a safe campus. The system detects and records the image of each motor vehicle of the past monitoring point in real time; the flow condition of the running vehicle is continuously and automatically recorded, analyzed and stored all the year round; the vehicle license plate recognition system has the advantages that vehicle license plate information is captured and recognized, the vehicle license plate recognition system is connected to a cloud platform and can be compared with a blacklist to capture, track and alarm suspicious vehicles, the recognition system firstly determines the optimal time period for vehicle operation and then uniformly recognizes each frame of image in the time period to obtain a comprehensive recognition result, recognition accuracy is greatly improved, each monitoring point recognizes the license plate, and therefore the suspicious vehicles can be tracked, and tracks of the vehicles are obtained.
The license plate analysis processing module also comprises a learning module and a license plate tracking module; the license plate tracking module feeds back a positioning result picture obtained by the license plate positioning module to the video to determine the motion track of the license plate in the video when the vehicle runs; the learning module is connected with the vehicle detection tracking module, the license plate positioning module, the license plate correction and fine positioning module, the license plate segmentation module, the character recognition module, the result decision module and the license plate tracking module, after the learning module analyzes and processes the license plates for multiple times, the learning module combines the motion track of the license plates in the video, the angle of the video, the picture quality and the vehicle running speed and the surrounding environment of the monitoring points to perform feedback guidance on each module, so that the license plate positioning module preferentially recognizes the position of the license plate track in each frame of picture and guides other modules to perform adaptive optimization adjustment according to the shooting angle and the shooting environment of the monitoring points, therefore, the recognition system can also have a self-learning mode and can adjust according to the environment adaptability of the monitoring points, and the recognition speed is improved.
In order to solve the second technical problem, the technical solution of the present invention is: a vehicle track tracking and license plate recognition method comprises the following steps:
A. setting a close-up camera at a monitoring point for shooting videos of passing vehicles;
B. analyzing videos shot by the close-up camera by using a vehicle detection and tracking module, and determining the optimal time period in the shot videos as a video analysis object according to the definition of a license plate and the size of a license plate display area after the vehicle detection and tracking module tracks the positions of the vehicles when the vehicle detection and tracking module finds that the vehicles pass through the shot videos;
C. the license plate positioning module takes a license plate frame as a distinguishing feature and combines a license plate ground color as an auxiliary distinguishing feature, analyzes each frame image in a video analysis object, divides each frame image in the video analysis object into an easy-to-position image and a difficult-to-position image, and the easy-to-position image comprises the distinguishing feature and the auxiliary distinguishing feature, so that the specific position of the license plate can be determined to obtain a positioning result image; when the license plate frame in the difficult positioning image is taken as a distinguishing technical feature and cannot be positioned, the auxiliary distinguishing feature of the license plate background color is used for re-positioning, the positioning result refers to the positioning result in other easy positioning images, if the license plate position of the positioning result in the difficult positioning image and the license plate position in the easy positioning image change according to a straight line track, the positioning result in the difficult positioning image is accurate, if the license plate position of the positioning result in the difficult positioning image and the license plate position of the easy positioning image do not change according to the straight line track, the positioning result of the difficult positioning image is inaccurate, and the easy positioning image without the positioning result is not taken as a judgment basis;
D. the license plate correction and fine positioning module corrects a positioning result picture and improves the sharpness and definition of the picture, a plurality of regions with the same color as the ground color of the license plate may exist in the positioning result picture of the image difficult to position, the license plate correction and fine positioning module screens the regions again and screens the regions again according to the running track of the vehicle to finally obtain a positioning result picture matched with the running track of the license plate in the image easy to position, and a group of fine positioning result pictures are obtained after the plurality of positioning result pictures are subjected to fine positioning;
E. dividing the fine positioning result picture into a plurality of pictures to be recognized containing single characters through a license plate segmentation module, and comparing each picture to be recognized with license plate characters stored in a database in advance through a character recognition module one by one to obtain a primary recognition result;
F. the result decision module analyzes all the preliminary recognition results, obtains the comprehensive credibility of the license plate according to the recording information of the license plate in the video process and the maximum credibility value of the preliminary recognition results, and finally obtains a conclusion of outputting the recognition result or refusing the result to continuously track the license plate;
preferably, the specific determination manner of the maximum confidence value in step F is: if the initial identification result is the only result character, the only result character is used as the final identification result, if a plurality of possible result characters are obtained in the initial identification result, the number of pictures of each possible result character obtained in the initial identification result is judged, the proportion of the easy-to-position pictures in the number of the pictures is referred to at the same time, and the possible result character which occupies the largest number of the pictures and has the largest easy-to-position pictures is used as the final identification result.
Preferably, the vehicle trajectory tracking and license plate recognition method further comprises a self-learning mode, and the specific mode is as follows: in videos shot by the close-up camera of each monitoring point, the shooting angle and static objects in the environment are always unchanged, the license plate tracking module is used for tracking the running track of the license plate in the videos, and after a plurality of video analysis objects are integrated to obtain a centralized running area of the license plate in a plurality of videos, the license plate positioning module and the license plate correcting and fine positioning module preferentially identify the centralized running area; and the license plate segmentation module and the character recognition module adaptively correct the segmentation angle and the character recognition angle after multiple recognition of the monitoring points.
Preferably, the close-up cameras in the vehicle track tracking and license plate recognition method comprise a head close-up camera for aligning with a head and a tail camera for aligning with a tail, the vehicle track tracking and license plate recognition method further comprises vehicle logo recognition, a common vehicle logo image of the vehicle, a head vehicle logo position image and a tail vehicle logo position image of each common vehicle type are stored in a database of the cloud computing platform in advance, the positioning vehicle logo is positioned while the vehicle license plate is positioned in the step C, and then the positioned vehicle logo is compared with the head vehicle logo position image and the tail vehicle logo position image in the database to obtain a vehicle logo recognition result.
After the technical scheme is adopted, the invention has the effects that: the identification method firstly determines the optimal time period as a video analysis object, thereby ensuring the accuracy of subsequent analysis and identification and reducing the identification difficulty; secondly, each frame of image of the video analysis object is identified and analyzed to obtain a comprehensive probabilistic result, so that the result is more accurate, and accidental probability is avoided; the recognition method corrects and precisely positions the license plate again after positioning, so that the recognition accuracy is further improved; finally, the recognition method is used for carrying out segmentation in a character segmentation mode and then carrying out independent comparison, and finally obtaining a recognition result with the maximum reliability by comprehensively judging the reliability of the recognition structure through a result decision module.
The specific determination method of the maximum confidence value in step F is as follows: if the preliminary identification result is the only result character, the only result character is used as the final identification result, if a plurality of possible result characters are obtained in the preliminary identification result, the number of pictures of the possible result characters obtained in the preliminary identification result is judged, the proportion of easily positioned pictures in the number of the pictures is referred to at the same time, the possible result character which occupies the largest number of the pictures and has the largest number of easily positioned pictures is used as the final identification result, and the judgment result accuracy of the maximum credibility value is higher.
In addition, the identification method can be used for carrying out adaptive adjustment according to the identification state of each detection point, so that the identification speed is improved, and the identification accuracy is not influenced. The identification method gives consideration to identification of the car logo, the license plate which can be further identified corresponds to the car logo, and the influence of the fake-licensed vehicle is reduced.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a functional diagram of an embodiment of the present invention;
in the drawings: 1. a video input management module; 11. close-up cameras; 12. a panoramic camera; 2. the license plate analysis processing module; 21. a vehicle detection tracking module; 22. a license plate positioning module; 23. the license plate correction and fine positioning module; 24. a license plate segmentation module; 25. a character recognition module; 26. a result decision module; 27. an output management module; 28. a learning module; 29. a license plate tracking module; 3. a downstream storage module.
Detailed Description
The present invention is described in further detail below with reference to specific examples.
As shown in fig. 1, a vehicle trajectory tracking and license plate recognition system includes close-up cameras 11 disposed at each monitoring point, the close-up cameras 11 are connected to downstream storage modules 3 disposed at each monitoring point through optical transceiver and encoder, the downstream storage modules 3 are in communication connection with a cloud computing platform, the cloud computing platform includes a license plate analysis processing module 2 dedicated to analyzing and recognizing license plates, of course, the cloud computing platform further includes a cloud storage module and other modules, and other smart systems in a city can be integrated on the cloud computing platform. In order to further facilitate manual observation, a panoramic camera 12 is additionally arranged on each monitoring point and is used for shooting the surrounding environment of the monitoring point at that time; and every camera all is provided with the high frequency flash light filling, makes things convenient for night to shoot. The close-up camera 11 and the panoramic camera 12 form a video input management module 1, and shot videos are transmitted to a downstream storage module 3 and are continuously stored in a mode of covering after capacity is full.
The monitoring points comprise entrance and exit monitoring points of roads, monitoring points of entrance and exit of each public place, and detection points of entrance and exit of a district and a school, wherein the number of the close-up cameras 11 on the monitoring points of the entrance and the exit of the roads corresponds to the number of the lanes of the roads one by one, wherein the head and the tail of the vehicle are respectively shot by the close-up cameras 11 arranged on the entrance and the exit monitoring points of the same road, the close-up cameras can be realized by adopting at least two modes, the installation directions of the close-up cameras 11 at the entrance and the close-up cameras 11 at the exit of the same road are different, for example, the tail is shot by the entrance camera, and the head is shot by the close-up cameras 11 at the exit. In another mode, a camera for shooting a vehicle head and a camera for shooting a vehicle tail are arranged at an inlet and a port, and both the two modes can be adopted, but the former is preferably adopted in the embodiment.
The license plate analysis processing module 2 comprises
The vehicle detection and tracking module 21 is used for analyzing the video shot by the close-up camera 11, and determining an optimal time period as a video analysis object according to the definition of a license plate and the size of a license plate display area after tracking the position of the vehicle; the vehicle detection tracking module 21 can well overcome various external interferences and obtain more reasonable identification results.
The license plate positioning module 22 is used for analyzing each frame image in the video analysis object by taking a license plate frame as a distinguishing characteristic and combining a license plate ground color comparison mode, and determining the specific position of the license plate to obtain a positioning result image; the method is suitable for various complex background environments and different camera angles, and the detection models of different license plate types can be trained quickly by learning enough samples because the camera angle of each detection point is unchanged. Wherein, foretell distinguishing characteristic need type in advance, the license plate of the vehicle of china has twenty many, promptly "large automobile front license plate, small-size car license plate, the automobile license plate of the container, the license plate of overseas car, the license plate of foreign nationality's car, test automobile license plate, train's license plate, trailer license plate, policeman's automobile license plate, police's small-size car license plate, army's large-size car license plate, the license plate of the embassy car, large-size car back license plate, 2002 formula license plate, agricultural transportation vehicle license plate, motorcycle license plate, tractor license plate, other license plates". Each license plate can be recorded with a pattern as a distinguishing characteristic, so that the license plates are convenient to identify.
And the license plate correction and fine positioning module 23 corrects the positioning result picture and improves the sharpness and definition of the image, then accurately positions the positioning result of each frame of image again, and removes irrelevant image parts again to obtain a fine positioning result picture.
The license plate segmentation module 24 segments the fine positioning result picture of each frame of image into a plurality of pictures to be recognized, and each picture to be recognized only contains one character; the license plate characters have various characteristics such as gray level, color, edge distribution and the like, can better inhibit the influence of other noises around the license plate, and can tolerate the license plate with a certain inclination angle. Various license plate common characters can be stored in a database of the cloud computing platform.
The character recognition module 25 is used for comparing each picture to be recognized with the license plate characters stored in the database one by one to obtain a primary recognition result; the pre-stored characters in the database include:
i is ten Arabic numerals of 0-9;
II twenty-six English letters A-Z;
III Chinese characters in province are abbreviated as (Jing, jin, Ji, Mongolia, Liao, Ji, Black, Shanghai, Su, Zhejiang, Wan, Min, gan, Lu, Yu, Hue, Xiang, Yue, Gui, Qiong, Chuan, Gui, Yu, Tibet, Shaanxi, Gan, Qing, Ning, Xin and Yu);
IV military brand Chinese characters (military, sea, air, Ji, North, Shen, south, orchid, Guang, Cheng, Ji);
v-number plate classification uses Chinese characters (police, study, collar, try, hang, harbor, Australia);
VI, number plate words of armed police;
a result decision module 26, wherein the result decision module 26 obtains the comprehensive reliability of the license plate according to the recording information of the license plate in the video process and the maximum reliability value of the primary recognition result obtained by combining each frame of image, and obtains a conclusion of outputting the recognition result or rejecting the result to continuously track the license plate; the recorded information comprises recognition results, recognition reliability, track records, similarity records and the like, and the final recognition result of one license plate is synthesized by analyzing the recognition results of all frames, intelligently classifying and voting the frames and combining certain grammar information. The method comprehensively utilizes the information of all frames, reduces accidental errors caused by the traditional identification algorithm based on a single image, and greatly improves the identification rate of the system and the correctness and reliability of the identification result.
And an output management module 27, wherein the output management module 27 outputs the recognition result obtained by the result decision module 26.
In this embodiment, the license plate analysis processing module 2 further includes a learning module 28 and a license plate tracking module 29;
the license plate tracking module 29 feeds back a positioning result picture obtained by the license plate positioning module 22 to the video to determine the motion track of the license plate in the video when the vehicle runs; the license plate tracking module 29 has a motion model and an update model with certain fault tolerance capability, so that license plates which are blocked for a short time or are blurred instantly can still be correctly tracked and predicted.
The learning module 28 is connected with the vehicle detection and tracking module 21, the license plate positioning module 22, the license plate correction and fine positioning module 23, the license plate segmentation module 24, the character recognition module 25, the result decision module 26 and the license plate tracking module 29, and after the learning module 28 analyzes and processes the license plates for multiple times, the learning module 28 feeds back and guides each module according to the motion track of the license plates in the video, the angle of the shot video, the image quality, the vehicle running speed and the surrounding environment of the monitoring point, so that the license plate positioning module 22 preferentially recognizes the position of the license plate track in each frame of image and guides other modules to perform adaptive optimization and adjustment according to the shooting angle and the shooting environment of the monitoring point.
The embodiment also discloses a vehicle track tracking and license plate recognition method, which comprises the following steps:
A. setting a close-up camera 11 at a monitoring point for shooting videos of passing vehicles;
B. analyzing the video shot by the close-up camera 11 by using the vehicle detection and tracking module 21, and when the vehicle detection and tracking module 21 finds that a vehicle passes through the shot video, determining the optimal time period in the shot video as a video analysis object according to the definition of a license plate and the size of a license plate display area after the vehicle detection and tracking module 21 tracks the position of the vehicle;
C. the license plate positioning module 22 takes a license plate frame as a distinguishing feature and combines a license plate ground color as an auxiliary distinguishing feature, the license plate positioning module 22 analyzes each frame of image in the video analysis object, divides each frame of image in the video analysis object into an easy-to-position image and a difficult-to-position image, and the easy-to-position image comprises the distinguishing feature and the auxiliary distinguishing feature, so that the specific position of the license plate can be determined to obtain a positioning result image; when the license plate frame in the difficult positioning image is taken as a distinguishing technical feature and cannot be positioned, the auxiliary distinguishing feature of the license plate background color is used for re-positioning, the positioning result refers to the positioning result in other easy positioning images, if the license plate position of the positioning result in the difficult positioning image and the license plate position in the easy positioning image change according to a straight line track, the positioning result in the difficult positioning image is accurate, if the license plate position of the positioning result in the difficult positioning image and the license plate position of the easy positioning image do not change according to the straight line track, the positioning result of the difficult positioning image is inaccurate, and the easy positioning image without the positioning result is not taken as a judgment basis;
D. the license plate correction and fine positioning module 23 corrects a positioning result picture and improves the sharpness and definition of the picture, a plurality of regions with the same color as the ground color of the license plate may exist in the positioning result picture of the image difficult to position, the license plate correction and fine positioning module 23 screens the regions again and screens the regions again according to the running track of the vehicle to finally obtain a positioning result picture matched with the running track of the license plate in the image easy to position, and a group of fine positioning result pictures are obtained after the plurality of positioning result pictures are subjected to fine positioning;
E. dividing the fine positioning result picture into a plurality of pictures to be recognized containing single characters through a license plate segmentation module 24, and comparing each picture to be recognized with the license plate characters stored in advance in a database one by one through a character recognition module 25 to obtain a primary recognition result;
F. the result decision module 26 analyzes all the preliminary recognition results, obtains the comprehensive credibility of the license plate according to the recording information of the license plate in the video process and the maximum credibility value of the preliminary recognition results, and finally obtains a conclusion of outputting the recognition result or refusing the result to continuously track the license plate;
the specific determination method of the maximum confidence value in step F is as follows: if the initial identification result is the only result character, the only result character is used as the final identification result, if a plurality of possible result characters are obtained in the initial identification result, the number of pictures of each possible result character obtained in the initial identification result is judged, the proportion of the easy-to-position pictures in the number of the pictures is referred to at the same time, and the possible result character which occupies the largest number of the pictures and has the largest easy-to-position pictures is used as the final identification result.
The vehicle track tracking and license plate recognition method further comprises a self-learning mode, and the specific mode is as follows: in the video shot by the close-up camera 11 of each monitoring point, the shot angle and the static objects in the environment are always unchanged, the license plate tracking module 29 is used for tracking the running track of the license plate in the video, and after a plurality of video analysis objects are integrated to obtain the centralized running area of the license plate in a plurality of videos, the license plate positioning module 22 and the license plate correction and fine positioning module 23 preferentially identify the centralized running area; the license plate segmentation module 24 and the character recognition module 25 adaptively correct the segmentation angle and the character recognition angle after multiple recognition of the monitoring point.
The close-up camera 11 in the vehicle track tracking and license plate recognition method comprises a head close-up camera 11 used for aligning the head to shoot, a tail camera used for aligning the tail to shoot, and the vehicle track tracking and license plate recognition method also comprises vehicle logo recognition, common vehicle logo images of the vehicle, head vehicle logo position images and tail vehicle logo position images of each type of common vehicle are stored in a database of a cloud computing platform in advance, a located vehicle logo is located while the vehicle logo is located in the step C, and then the located vehicle logo is compared with the head vehicle logo position images and the tail vehicle logo position images in the database to obtain a vehicle logo recognition result. The identification of the emblem is not essential and may be activated on an actual basis, for example, when a vehicle is suspected of being a fake plate, the function may be activated.
The above-mentioned embodiments are merely descriptions of the preferred embodiments of the present invention, and are not intended to limit the scope of the present invention, and various modifications and alterations made to the technical solution of the present invention without departing from the spirit of the present invention are intended to fall within the scope of the present invention defined by the claims.
Claims (4)
1. A vehicle track tracking and license plate recognition method is characterized in that: the method uses a vehicle track tracking and license plate recognition system, the system comprises close-up cameras arranged on each monitoring point, the close-up cameras are connected with downstream storage modules arranged on each monitoring point through optical transceivers and encoders, the downstream storage modules are in communication connection with a cloud computing platform, the cloud computing platform comprises a license plate analysis processing module special for analyzing and recognizing license plates, and the license plate analysis processing module comprises a license plate analysis processing module special for analyzing and recognizing the license plates
The vehicle detection tracking module is used for analyzing the video shot by the close-up camera, and determining the optimal time period as a video analysis object according to the definition of the license plate and the size of the license plate display area after tracking the position of the vehicle;
the license plate positioning module analyzes each frame image in the video analysis object in a mode of taking a license plate frame as a distinguishing characteristic and combining license plate ground color comparison, and determines the specific position of the license plate to obtain a positioning result image;
the license plate correction and fine positioning module corrects the positioning result picture and improves the sharpness and definition of the image, then accurately positions the positioning result of each frame of image, and removes irrelevant image parts again to obtain a fine positioning result picture;
the license plate segmentation module segments the fine positioning result picture of each frame of image into a plurality of pictures to be recognized, and each picture to be recognized only contains one character;
the character recognition module compares each picture to be recognized with the license plate characters stored in the database in advance one by one to obtain a primary recognition result;
the result decision module is used for obtaining the comprehensive reliability of the license plate according to the recording information of the license plate in the video process and the maximum reliability value of the primary recognition result obtained by combining each frame of picture, and obtaining a conclusion of outputting the recognition result or refusing the result to continuously track the license plate;
the output management module outputs the identification result obtained by the result decision module;
the method comprises the following steps:
A. setting a close-up camera at a monitoring point for shooting videos of passing vehicles;
B. analyzing videos shot by the close-up camera by using a vehicle detection and tracking module, and determining the optimal time period in the shot videos as a video analysis object according to the definition of a license plate and the size of a license plate display area after the vehicle detection and tracking module tracks the positions of the vehicles when the vehicle detection and tracking module finds that the vehicles pass through the shot videos;
C. the license plate positioning module takes a license plate frame as a distinguishing feature and combines a license plate ground color as an auxiliary distinguishing feature, analyzes each frame image in a video analysis object, divides each frame image in the video analysis object into an easy-to-position image and a difficult-to-position image, and the easy-to-position image comprises the distinguishing feature and the auxiliary distinguishing feature, so that the specific position of the license plate can be determined to obtain a positioning result image; when the license plate frame in the difficult positioning image is taken as a distinguishing technical feature and cannot be positioned, the auxiliary distinguishing feature of the license plate background color is used for re-positioning, the positioning result refers to the positioning result in other easy positioning images, if the license plate position of the positioning result in the difficult positioning image and the license plate position in the easy positioning image change according to a straight line track, the positioning result in the difficult positioning image is accurate, if the license plate position of the positioning result in the difficult positioning image and the license plate position of the easy positioning image do not change according to the straight line track, the positioning result of the difficult positioning image is inaccurate, and the easy positioning image without the positioning result is not taken as a judgment basis;
D. the license plate correction and fine positioning module corrects a positioning result picture and improves the sharpness and definition of the picture, a plurality of regions with the same color as the ground color of the license plate may exist in the positioning result picture of the image difficult to position, the license plate correction and fine positioning module screens the regions again and screens the regions again according to the running track of the vehicle to finally obtain a positioning result picture matched with the running track of the license plate in the image easy to position, and a group of fine positioning result pictures are obtained after the plurality of positioning result pictures are subjected to fine positioning;
E. dividing the fine positioning result picture into a plurality of pictures to be recognized containing single characters through a license plate segmentation module, and comparing each picture to be recognized with license plate characters stored in a database in advance through a character recognition module one by one to obtain a primary recognition result;
F. and the result decision module analyzes all the preliminary recognition results, obtains the comprehensive credibility of the license plate according to the recorded information of the license plate in the video process and the maximum credibility value of the preliminary recognition results, and finally obtains a conclusion of outputting the recognition result or refusing the result to continuously track the license plate.
2. The method for tracking a vehicle trajectory and recognizing a license plate of claim 1, wherein: the specific determination method of the maximum confidence value in step F is as follows: if the initial identification result is the only result character, the only result character is used as the final identification result, if a plurality of possible result characters are obtained in the initial identification result, the number of pictures of each possible result character obtained in the initial identification result is judged, the proportion of the easy-to-position pictures in the number of the pictures is referred to at the same time, and the possible result character which occupies the largest number of the pictures and has the largest easy-to-position pictures is used as the final identification result.
3. The method for tracking a vehicle trajectory and recognizing a license plate of claim 2, wherein: the vehicle track tracking and license plate recognition method further comprises a self-learning mode, and the specific mode is as follows: in videos shot by the close-up camera of each monitoring point, the shooting angle and static objects in the environment are always unchanged, the license plate tracking module is used for tracking the running track of the license plate in the videos, and after a plurality of video analysis objects are integrated to obtain a centralized running area of the license plate in a plurality of videos, the license plate positioning module and the license plate correcting and fine positioning module preferentially identify the centralized running area; and the license plate segmentation module and the character recognition module adaptively correct the segmentation angle and the character recognition angle after multiple recognition of the monitoring points.
4. The method for tracking a vehicle trajectory and recognizing a license plate of claim 3, wherein: the close-up camera in the vehicle track tracking and license plate recognition method comprises a head close-up camera used for aligning the head to shoot, a tail camera used for aligning the tail to shoot, and the vehicle track tracking and license plate recognition method also comprises vehicle logo recognition, common vehicle logo images of the vehicle, head vehicle logo position images and tail vehicle logo position images of each type of common vehicle types are stored in a database of a cloud computing platform in advance, a positioned vehicle logo is positioned while the vehicle logo is positioned in the step C, and then the positioned vehicle logo is compared with the head vehicle logo position images and the tail vehicle logo position images in the database to obtain a vehicle logo recognition result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611216113.0A CN106778648B (en) | 2016-12-26 | 2016-12-26 | Vehicle track tracking and license plate recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611216113.0A CN106778648B (en) | 2016-12-26 | 2016-12-26 | Vehicle track tracking and license plate recognition method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106778648A CN106778648A (en) | 2017-05-31 |
| CN106778648B true CN106778648B (en) | 2020-02-07 |
Family
ID=58927403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611216113.0A Active CN106778648B (en) | 2016-12-26 | 2016-12-26 | Vehicle track tracking and license plate recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106778648B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109801498A (en) * | 2017-11-16 | 2019-05-24 | 北京筑梦园科技有限公司 | Licence plate recognition methods a kind of while using a variety of triggering modes |
| CN109712428A (en) * | 2019-02-27 | 2019-05-03 | 爱笔（北京）智能科技有限公司 | A kind of parking position localization method and device, system |
| CN110164136B (en) * | 2019-03-29 | 2021-05-04 | 重庆市城投金卡信息产业(集团)股份有限公司 | Fake-licensed vehicle identification method |
| CN110491133B (en) * | 2019-08-08 | 2020-10-16 | 善泊科技(珠海)有限公司 | Vehicle information correction system and method based on confidence |
| CN110991255B (en) * | 2019-11-11 | 2023-09-08 | 智慧互通科技股份有限公司 | Method for detecting fake-licensed car based on deep learning algorithm |
| CN110807934A (en) * | 2019-11-13 | 2020-02-18 | 赵洪涛 | Intelligent scheduling platform for monitoring highway operation |
| CN111666860A (en) * | 2020-06-01 | 2020-09-15 | 浙江省机电设计研究院有限公司 | Vehicle track tracking method integrating license plate information and vehicle characteristics |
| CN112183528B (en) * | 2020-09-23 | 2022-06-21 | 桂林电子科技大学 | Method for tracking target vehicle, device, system and computer storage medium thereof |
| CN112489436B (en) * | 2020-10-29 | 2022-05-03 | 浙江预策科技有限公司 | Vehicle identity recognition method, device and system and electronic device |
| CN112309126B (en) * | 2020-10-30 | 2022-07-05 | 杭州海康威视数字技术股份有限公司 | License plate detection method and device, electronic equipment and computer readable storage medium |
| CN112464796B (en) * | 2020-11-25 | 2024-05-24 | 迪蒙智慧交通科技有限公司 | Vehicle searching method, vehicle searching system and computer readable storage medium |
| CN112836683B (en) * | 2021-03-04 | 2024-07-09 | 广东建邦计算机软件股份有限公司 | License plate recognition method, device, equipment and medium for portable camera equipment |
| CN113469101A (en) * | 2021-07-13 | 2021-10-01 | 广联达科技股份有限公司 | Vehicle weight recognition method, device, equipment and readable storage medium |
| CN113657378B (en) * | 2021-07-28 | 2024-04-26 | 讯飞智元信息科技有限公司 | Vehicle tracking method, vehicle tracking system and computing device |
| CN114722309A (en) * | 2022-03-30 | 2022-07-08 | 广东艾科智泊科技股份有限公司 | License plate recognition method based on panoramic parking lot |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201159903Y (en) * | 2008-03-14 | 2008-12-03 | 北京信路威科技发展有限公司 | License plate recognition device |
| CN101630361A (en) * | 2008-12-30 | 2010-01-20 | 北京邮电大学 | Plate number, body color and mark identification-based equipment and plate number, body color and mark identification-based method for identifying fake plate vehicles |
| CN102722704A (en) * | 2012-06-12 | 2012-10-10 | 厦门宸天电子科技有限公司 | Method and system for recognizing vehicle license plate by integrating video dynamic tracking |
| CN105893909A (en) * | 2014-10-16 | 2016-08-24 | 南京索酷信息科技有限公司 | Method for applying license plate recognition to intelligent community |
-
2016
- 2016-12-26 CN CN201611216113.0A patent/CN106778648B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201159903Y (en) * | 2008-03-14 | 2008-12-03 | 北京信路威科技发展有限公司 | License plate recognition device |
| CN101630361A (en) * | 2008-12-30 | 2010-01-20 | 北京邮电大学 | Plate number, body color and mark identification-based equipment and plate number, body color and mark identification-based method for identifying fake plate vehicles |
| CN102722704A (en) * | 2012-06-12 | 2012-10-10 | 厦门宸天电子科技有限公司 | Method and system for recognizing vehicle license plate by integrating video dynamic tracking |
| CN105893909A (en) * | 2014-10-16 | 2016-08-24 | 南京索酷信息科技有限公司 | Method for applying license plate recognition to intelligent community |
Non-Patent Citations (3)
| Title |
|---|
| 《vehicle munber recognition system for automatic toll tax collection》;Shoaib Rehman Soomro et al;《2012 international conference of robotics and artificial intelligence》;20121023;全文 * |
| 《基于神经网络的视频序列车牌定位算法》;程晓航等;《计算机工程与设计》;20070930;全文 * |
| 《基于视频检测的车牌照定位方法》;聂艳红等;《计算机工程与设计》;20070831;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106778648A (en) | 2017-05-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106778648B (en) | Vehicle track tracking and license plate recognition method | |
| US8995723B2 (en) | Detecting and recognizing traffic signs | |
| CN104778444B (en) | The appearance features analysis method of vehicle image under road scene | |
| CN110969160B (en) | License plate image correction and recognition method and system based on deep learning | |
| US20130287261A1 (en) | Electronic device and method for managing traffic flow | |
| TW202013252A (en) | License plate recognition system and license plate recognition method | |
| CN105788286A (en) | Intelligent red light running identifying system and vehicle behavior detecting and capturing method | |
| MX2010005149A (en) | Security systems. | |
| CN111860352B (en) | Multi-lens vehicle track full tracking system and method | |
| KR101756848B1 (en) | Unlawfulness parking and no standing control system and method thereof | |
| CN111382735B (en) | Night vehicle detection method, device, equipment and storage medium | |
| US20220405514A1 (en) | Techniques for capturing enhanced images for pattern indentifications | |
| US11978260B2 (en) | Systems and methods for rapid license plate reading | |
| AU2021240327A1 (en) | Infringement detection method, device and system | |
| US8229170B2 (en) | Method and system for detecting a signal structure from a moving video platform | |
| CN106571040A (en) | Suspicious vehicle confirmation method and equipment | |
| WO2022271468A1 (en) | Techniques for improving an image readability using one or more patterns | |
| CN108182393A (en) | A kind of automobile and its front truck tracking and system of application | |
| CN111507196A (en) | Vehicle type identification method based on machine vision and deep learning | |
| US10417518B2 (en) | Vehicle camera system | |
| Yim et al. | Integrated plate recognition and speed detection for intelligent transportation systems | |
| CN108985197B (en) | Automatic detection method for taxi driver smoking behavior based on multi-algorithm fusion | |
| CN105427615A (en) | Robust red-light-running snapshotting system and method under low illumination | |
| KR101547239B1 (en) | System and method for adjusting camera brightness based extraction of background image | |
| CN111881843B (en) | Face detection-based taxi passenger carrying number counting method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 215600 Zhangjiagang Economic Development Zone, Suzhou City, Jiangsu Province (No. 6 Xinjing West Road, South District) Jiangsu Shengshi Huaan Intelligent Technology Co., Ltd. Applicant after: Jiangsu Shengshi Huaan Internet Technology Co.,Ltd. Address before: 215600 Zhangjiagang Economic Development Zone, Suzhou City, Jiangsu Province (No. 6 Xinjing West Road, South District) Jiangsu Shengshi Huaan Intelligent Technology Co., Ltd. Applicant before: JIANGSU SHENGSHI HUAAN INTELLIGENT TECHNOLOGY CO.,LTD. |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200402 Address after: 215600, No. 10, Jing Jing Road, Yang family house, Zhangjiagang, Jiangsu, Suzhou Patentee after: ZHANGJIAGANG XINDONGPO GLASS COTTON PRODUCTS FACTORY Address before: 215600, Zhangjiagang Economic Development Zone, Suzhou, Jiangsu province (6 West Jingyang Road), Jiangsu Shengshi Huaan intelligent Polytron Technologies Inc Patentee before: Jiangsu Shengshi Huaan Internet Technology Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 215600 No. 10, Chang Jing Road, Yang She Town, Zhangjiagang City, Suzhou, Jiangsu Patentee after: Suzhou Xingheng Electronic Energy Saving Technology Co.,Ltd. Address before: 215600 No. 10, Chang Jing Road, Yang She Town, Zhangjiagang City, Suzhou, Jiangsu Patentee before: ZHANGJIAGANG XINDONGPO GLASS COTTON PRODUCTS FACTORY |
|
| CP01 | Change in the name or title of a patent holder | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230601 Address after: 215600 B2501, 22 Zhongxing Middle Road, Tongxin building, yangshe Town, Zhangjiagang City, Suzhou City, Jiangsu Province Patentee after: Suzhou Dongfang jinhuitong Intelligent Technology Co.,Ltd. Address before: 215600 No. 10, Chang Jing Road, Yang She Town, Zhangjiagang City, Suzhou, Jiangsu Patentee before: Suzhou Xingheng Electronic Energy Saving Technology Co.,Ltd. |
|
| TR01 | Transfer of patent right |