CN111508269A - Open type parking space vehicle distinguishing method and device based on image recognition - Google Patents

Open type parking space vehicle distinguishing method and device based on image recognition Download PDF

Info

Publication number
CN111508269A
CN111508269A CN202010331657.1A CN202010331657A CN111508269A CN 111508269 A CN111508269 A CN 111508269A CN 202010331657 A CN202010331657 A CN 202010331657A CN 111508269 A CN111508269 A CN 111508269A
Authority
CN
China
Prior art keywords
license plate
picture
vehicle
detected
exists
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010331657.1A
Other languages
Chinese (zh)
Other versions
CN111508269B (en
Inventor
鲁继勇
王海峰
赖胜军
韩道猛
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.)
Shenzhen Zhiyouting Technology Co ltd
Original Assignee
Shenzhen Zhiyouting Technology Co ltd
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 Shenzhen Zhiyouting Technology Co ltd filed Critical Shenzhen Zhiyouting Technology Co ltd
Priority to CN202010331657.1A priority Critical patent/CN111508269B/en
Publication of CN111508269A publication Critical patent/CN111508269A/en
Application granted granted Critical
Publication of CN111508269B publication Critical patent/CN111508269B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/14—Traffic control systems for road vehicles indicating individual free spaces in parking areas
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • 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
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
    • G08G1/0175—Detecting movement of traffic to be counted or controlled identifying vehicles by photographing vehicles, e.g. when violating traffic rules
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/14—Traffic control systems for road vehicles indicating individual free spaces in parking areas
    • G08G1/145—Traffic control systems for road vehicles indicating individual free spaces in parking areas where the indication depends on the parking areas
    • G08G1/148—Management of a network of parking areas
    • 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)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Molecular Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Multimedia (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an open type parking space vehicle distinguishing method based on image recognition, which comprises the following steps: s1, acquiring a picture to be detected; s2, detecting the license plate of the picture to be detected by using a license plate detection algorithm, judging whether the license plate exists in the picture, outputting license plate information of the picture when the license plate exists, and inputting the picture into the step S3 for processing when the license plate does not exist; and S3, judging whether the car exists or not in the picture, if the car exists, outputting the picture that the car exists, and if the car does not exist, outputting the picture that the car does not exist. The method and the device provided by the invention can effectively improve the accuracy of vehicle identification in the open parking space, can provide friendly service for users, and effectively improve the user experience.

Description

Open type parking space vehicle distinguishing method and device based on image recognition
Technical Field
The invention relates to an open type parking space vehicle distinguishing method and device based on image recognition.
Background
The existing parking management systems are mostly managed in districts, in commercial venues, on-ground and underground closed places, vehicles can be easily managed in the closed parking lots, and the vehicle plate recognition cameras and the management systems are arranged at entrances and exits, so that the time of vehicle access and payment can be accurately managed. The invention discloses a road side parking management system, which is characterized in that a road side is an open scene, so that no effective means is provided for managing parking spaces for road side parking, a top-mounted camera is adopted at present, but the top-mounted camera is troublesome in construction, expensive and poor in usability.
Disclosure of Invention
Aiming at the current technical situation, the invention provides an open type parking space vehicle distinguishing method based on image recognition, which comprises the following steps:
s1, acquiring a picture to be detected, capturing a vehicle through a camera in parking lot parking space management equipment installed on the basis of the road side in an open parking lot, and acquiring the picture to be detected;
s2, detecting the license plate of the picture to be detected by using a license plate detection algorithm, judging whether the license plate exists in the picture, outputting license plate information of the picture when the license plate exists, and inputting the picture into the step S3 for processing when the license plate does not exist;
and S3, judging whether the car exists or not in the picture, if the car exists, outputting the picture that the car exists, and if the car does not exist, outputting the picture that the car does not exist.
In addition, the invention also provides an open type parking space vehicle distinguishing device based on image recognition, which comprises:
the image acquisition module to be detected captures a vehicle through a camera in parking lot parking space management equipment installed on the basis of the road side in an open parking lot to acquire an image to be detected;
the license plate detection module is used for detecting the license plate of the picture to be detected by using a license plate detection algorithm, judging whether the license plate exists in the picture or not, outputting license plate information of the picture when the license plate exists, and inputting the picture into the step S3 for processing when the license plate does not exist;
and the vehicle-presence and vehicle-absence classification module is used for performing vehicle-presence and vehicle-absence classification judgment on the picture, outputting the picture with a vehicle if the picture has a vehicle, and outputting the picture without a vehicle if the picture does not have a vehicle.
According to the technical scheme, the vehicles in the open parking lot are identified, the states of the vehicles are monitored, the accuracy of vehicle identification in the open parking lot is improved, friendly services can be provided for users, and the use experience of the users is effectively improved.
Drawings
Fig. 1 is a schematic flowchart of an open parking space vehicle identification method based on image recognition according to an embodiment of the present application;
FIG. 2 is a schematic view of an open parking space according to an embodiment of the present application;
FIG. 3 is a schematic diagram of the fast-RCNN algorithm model in an embodiment of the present application;
FIG. 4 is a schematic view of an SSD algorithm model in an embodiment of the present application;
FIG. 5 is a schematic view of the Yolo-v3 model in an embodiment of the present application;
FIG. 6 is a schematic diagram of a residual error network element in an embodiment of the present application;
FIG. 7-1 is an exemplary illustration of exposed wheel categories in the classification of vehicle present and absent in an exemplary embodiment of the present application;
FIG. 7-2 is an exemplary illustration of exposed chassis categories in the classification of vehicle and non-vehicle in the exemplary embodiment of the present application;
FIGS. 7-3 are exemplary diagrams of wheel-only categories in the classification of vehicle-on-vehicle and vehicle-off in accordance with embodiments of the present disclosure;
FIGS. 7-4 are exemplary diagrams of classes of vehicles in a classification of vehicles with or without vehicles in the embodiment of the present application;
FIGS. 7-5 are exemplary illustrations of remote vehicle categories in the classification of vehicle and non-vehicle in the exemplary embodiment of the present application;
FIGS. 7-6 are exemplary diagrams of exposure intensity categories in the classification of vehicle or non-vehicle in the embodiment of the present application;
fig. 8 is a schematic diagram of a mobilenet network structure in the embodiment of the present application.
Detailed Description
The invention will now be described in detail with reference to the preferred embodiments thereof.
Example one
The embodiment provides an open parking space vehicle identification method based on image recognition, as shown in fig. 1, the method includes:
and S1, acquiring the picture to be detected.
The method is applied to an open parking lot, as shown in fig. 2, a parking lot parking space management device including an ultrasonic detector and a camera is generally installed at a position 18 cm away from a frame on a parking space (which can be adjusted according to the length of an actual parking space), the ultrasonic detection distance generally covers half of the parking space, the ultrasonic detector regularly transmits ultrasonic waves to detect the distance of an object in the parking space, the transmission time is reduced according to echo time to calculate the distance, vehicles are counted when the distance is generally less than 5 meters and greater than 0.2 meter, the ultrasonic waves are used for judging whether the parking space management device is started, vehicle image capturing is started through the camera after the device is started, and a group of captured images is identified.
And S2, detecting the license plate of the picture to be detected by using a license plate detection algorithm, judging whether the license plate exists in the picture, outputting license plate information of the picture when the license plate exists, and inputting the picture into the step S3 for processing when the license plate does not exist.
Preferably, the step of detecting the license plate of the picture to be detected by using a license plate detection algorithm and judging whether the license plate exists in the picture further comprises the following steps:
s2.1, obtaining a license plate region in the image by using a license plate detection algorithm, wherein the license plate detection algorithm can adopt, but is not limited to, algorithms such as fast-RCNN, SSD, yolov-v1, yolo-v2 and yolo-v3 to carry out license plate detection.
The fast-RCNN algorithm is an upgrading algorithm of RCNN and fast-RCNN, a model is shown in FIG. 3, and the specific flow is as follows:
1) inputting the whole picture into a convolutional neural network CNN to obtain a convolutional feature map;
2) inputting the convolution characteristic into an RPN (region pro-social network) to obtain characteristic information of the candidate frame;
3) judging whether the feature information extracted from the candidate frame belongs to a specific class by using a classifier;
4) and further adjusting the position of the candidate frame belonging to a certain characteristic by using a regressor.
By adopting the SSD algorithm, the SSD model is as shown in fig. 4, and the main flow of license plate recognition is as follows:
1) inputting a picture, extracting features of the picture through a Convolutional Neural Network (CNN), and generating a convolutional featuremap;
2) extracting feature maps of six layers, and then generating a defaultbox at each point of the feature maps, wherein the number of each layer is different, but each point is provided with the same;
3) and collecting all the generated default boxes, totally discarding the default boxes into NMS (non-maximum suppression method), and outputting the filtered default boxes and outputting.
The SSD algorithm is a great improvement in three areas:
multi-scale: the SSD detects targets of different scales using 6 different feature maps, a low-level predicts small targets, and a high-level predicts large targets, so that smaller and larger targets can be detected at the same time.
Various aspect ratio default boxes (anchors boxes) are provided: at each pixel point of the feature map, default box (anchor box) with different aspect ratio is generated, and the aspect ratio can be set to be {1, 2, 3, 1/2, 1/3 }. Assuming that each pixel point has k default boxes, each default box needs to be classified and regressed, wherein the number of convolution kernels used for classification is ck (c represents the number of categories), and the number of convolution kernels used for regression is 4 k.
Data enhancement: two ways of data enhancement are used in SSDs. 1) And (3) amplification operation: the IOU of the random crop, patch and any object is 0.1, 0.3, 0.5, 0.7 and 0.9, the size of each patch is [0.1, 1] of the size of the original image, the aspect ratio is between 1/2 and 2, and more objects with larger scale can be generated; 2) and (3) reducing operation: first create a canvas 16 times the size of the original, then place the original in it, then randomly crop, which can generate more smaller scale objects.
The primary flow of the Yolo-v1 algorithm for license plate detection is as follows:
1) the original picture is scaled to 448 x 448, which is to be convenient for the following integer division.
2) The picture is divided into s-by-s grids, and if the center of an object falls on a certain cell, the cell is responsible for predicting the object. B bounding box values need to be predicted for each cell, with a confidence predicted for each bounding box. This confidence is not just the probability that the bounding box is the object to be detected, which is the product of the probability of the object to be detected times the bounding box multiplied by the true position IoU, thereby reflecting the accuracy of the predicted position of the bounding box.
Because the Yolo-v1 is not very good for positioning the bounding box, and has a certain difference in precision compared with the similar network, the Yolo-v2 algorithm is proposed:
YO L Ov2 greatly changes speed and precision and absorbs the advantages of the networks of the same type, and tries in steps, YO L0 v2 is provided after improvement on the basis of v1, is inspired by faster RCNN, introduces anchors, uses a k-means method at the same time, discusses the number of anchors, makes a compromise between precision and speed, modifies a network structure, removes a full connection layer, and changes the full convolution structure.
The Yolo-v 3: Yolo-v3 is a target detection network which is most balanced in speed and precision so far, short plates of YO L O series are completely supplemented with the small objects which are high in speed and not good for detection through fusion of various advanced methods, and a specific model is shown in FIG. 5.
Each box uses multi-label classification to predict classes that the bounding box may contain. Class prediction is performed using a binary cross entropy penalty, without using a softmax classifier. For overlapping tags, the multi-tag approach may better simulate data.
And predicting across sizes. And by adopting an up-sampling and fusion method similar to FPN (field programmable gate array), detection is performed on the feature maps of a plurality of scales, and the detection effect on small targets is improved.
And S2.2, license plate classification, namely cutting off the detected license plate region, and performing secondary classification by adopting a convolution network model to distinguish whether the license plate region is a license plate.
Because some pseudo regions exist in the detected license plate region, the invention cuts the detected license plate region, adopts a convolution network model to carry out secondary classification, and distinguishes whether the license plate is the license plate or not, and can adopt but not limited to a deepface model, a resnet model and the like.
The Resnet model is obtained by introducing a residual error unit shown in FIG. 6 into a cnn convolutional neural network, and the Resnet model can enable a stack layer of the network to learn new features on the basis of input features, so that the network has better performance.
And S3, judging whether the car exists or not in the picture, if the car exists, outputting the picture that the car exists, and if the car does not exist, outputting the picture that the car does not exist.
The image classification model is used for classifying the images, and the classification judgment of whether the vehicles exist in the images or not is completed. Specifically, the scene images may be classified by using a convolutional neural network, such as alexnet, surface, mobilnet, resnet models, and the like, and the classification may be classified into six classes according to the types of the images, as shown in fig. 7-1 to 7-6: a vehicle sample: vehicle body wheels, only chassis (no wheels), only wheels. No vehicle sample: there is no car in the scene, there is a passing car (remote car) in the scene, and the picture is exposed.
The general configuration of the mobilenet network is shown in fig. 8.
Example two
The embodiment of the present application further includes an open parking space vehicle discriminating device based on image recognition, including:
the device is applied to an open parking lot, as shown in fig. 2, parking lot parking space management equipment comprising an ultrasonic detector and a camera is generally installed at a position 18 cm away from a frame on a parking space (can be adjusted according to the length of an actual parking space), the ultrasonic detection distance generally covers half of the parking space, the ultrasonic detector regularly transmits ultrasonic waves for detecting the distance of objects in the parking space, the transmission time is reduced according to echo time, the distance is calculated, vehicles are generally calculated when the distance is less than 5 m and more than 0.2 m, the ultrasonic waves are used for judging whether parking space management equipment is started, vehicle image snapshot is started through the camera after the equipment is started, and a group of captured images are identified as the to-be-detected images.
The license plate detection module: and detecting the license plate of the picture to be detected by using a license plate detection algorithm, judging whether the license plate exists in the picture, outputting license plate information of the picture when the license plate exists, and inputting the picture into a vehicle-presence or vehicle-absence classification module for processing when the license plate does not exist.
Preferably, the step of detecting the license plate of the picture to be detected by using a license plate detection algorithm and judging whether the license plate exists in the picture further comprises the following steps:
and the license plate region acquisition sub-module acquires a license plate region in the image by using a license plate detection algorithm, wherein the license plate detection algorithm can adopt, but is not limited to, algorithms such as fast-RCNN, SSD, yolov-v1, yolo-v2 and yolo-v3 to carry out license plate detection.
The fast-RCNN algorithm is an upgrading algorithm of RCNN and fast-RCNN, a model is shown in FIG. 3, and the specific flow is as follows:
1) inputting the whole picture into a convolutional neural network CNN to obtain a convolutional feature map;
2) inputting the convolution characteristic into an RPN (region pro-social network) to obtain characteristic information of the candidate frame;
3) judging whether the feature information extracted from the candidate frame belongs to a specific class by using a classifier;
4) and further adjusting the position of the candidate frame belonging to a certain characteristic by using a regressor.
By adopting the SSD algorithm, the SSD model is as shown in fig. 4, and the main flow of license plate recognition is as follows:
1) inputting a picture, extracting features of the picture through a Convolutional Neural Network (CNN), and generating a convolutional featuremap;
2) extracting feature maps of six layers, and then generating a defaultbox at each point of the feature maps, wherein the number of each layer is different, but each point is provided with the default map;
3) and collecting all the generated default boxes, totally discarding the default boxes into NMS (non-maximum suppression method), and outputting the filtered default boxes and outputting.
The SSD algorithm is a great improvement in three areas:
multi-scale: the SSD detects targets of different scales using 6 different feature maps, a low-level predicts small targets, and a high-level predicts large targets, so that smaller and larger targets can be detected at the same time.
Various aspect ratio default boxes (anchors boxes) are provided: at each pixel point of the feature map, default box (anchor box) with different aspect ratio is generated, and the aspect ratio can be set to be {1, 2, 3, 1/2, 1/3 }. Assuming that each pixel point has k defaultbox, each defaultbox needs to be classified and regressed, wherein the number of convolution kernels used for classification is ck (c represents the number of categories), and the number of convolution kernels used for regression is 4 k.
Data enhancement: two ways of data enhancement are used in SSDs. 1) And (3) amplification operation: the IOU of the random crop, patch and any object is 0.1, 0.3, 0.5, 0.7 and 0.9, the size of each patch is [0.1, 1] of the size of the original image, the aspect ratio is between 1/2 and 2, and more objects with larger scale can be generated; 2) and (3) reducing operation: first create a canvas 16 times the size of the original, then place the original in it, then randomly crop, which can generate more smaller scale objects.
The primary flow of the Yolo-v1 algorithm for license plate detection is as follows:
1) the original picture is scaled to 448 x 448, which is to be convenient for the following integer division.
2) The picture is divided into s-by-s grids, and if the center of an object falls on a certain cell, the cell is responsible for predicting the object. B bounding box values need to be predicted for each cell, with a confidence predicted for each bounding box. This confidence is not just the probability that the bounding box is the object to be detected, which is the product of the probability of the object to be detected times the bounding box multiplied by the true position IoU, thereby reflecting the accuracy of the predicted position of the bounding box.
Because the Yolo-v1 is not very good for positioning the bounding box, and has a certain difference in precision compared with the similar network, the Yolo-v2 algorithm is proposed:
YO L Ov2 greatly changes speed and precision, and absorbs the advantages of the networks of the same type, and tries in steps, YO L Ov2 is provided after improvement on the basis of v1, is inspired by faster RCNN, introduces anchors, uses a k-means method at the same time, discusses the number of anchors, makes a compromise between precision and speed, modifies a network structure, removes a full connection layer, and changes the network structure into a full convolution structure.
The Yolo-v 3: Yolo-v3 is a target detection network which is most balanced in speed and precision so far, short plates of YO L O series are completely supplemented with the small objects which are high in speed and not good for detection through fusion of various advanced methods, and a specific model is shown in FIG. 5.
Each box uses multi-label classification to predict classes that the bounding box may contain. Class prediction is performed using a binary cross entropy penalty, without using a softmax classifier. For overlapping tags, the multi-tag approach may better simulate data.
And predicting across sizes. And by adopting an up-sampling and fusion method similar to FPN (field programmable gate array), detection is performed on the feature maps of a plurality of scales, and the detection effect on small targets is improved.
And the license plate classification submodule cuts the detected license plate area, performs secondary classification by adopting a convolution network model, and distinguishes whether the license plate area is a license plate or not.
Because some pseudo regions exist in the detected license plate region, the invention cuts the detected license plate region, adopts a convolution network model to carry out secondary classification, and distinguishes whether the license plate is the license plate or not, and can adopt but not limited to a deepface model, a resnet model and the like.
The Resnet model is obtained by introducing a residual error unit shown in FIG. 6 into a cnn convolutional neural network, and the Resnet model can enable a stack layer of the network to learn new features on the basis of input features, so that the network has better performance.
The classification module for the vehicle with or without the vehicle: and (4) judging whether the vehicle exists or not in the picture in a classified manner, if the vehicle exists, outputting the picture that the vehicle exists, and if the vehicle does not exist, outputting the picture that the vehicle does not exist.
The image classification model is used for classifying the images, and the classification judgment of whether the vehicles exist in the images or not is completed. Specifically, the scene images may be classified by using a convolutional neural network, such as alexnet, surface, mobilnet, resnet models, and the like, and the classification may be classified into six classes according to the types of the images, as shown in fig. 7-1 to 7-6: a vehicle sample: vehicle body wheels, only chassis (no wheels), only wheels. No vehicle sample: there is no car in the scene, there is a passing car (remote car) in the scene, and the picture is exposed.
The general configuration of the mobilenet network is shown in fig. 8.
According to the technical scheme, the vehicles in the open parking lot are identified, the states of the vehicles are monitored, the accuracy of vehicle identification in the open parking lot is improved, friendly services can be provided for users, and the use experience of the users is effectively improved.
The technical solution of the embodiment of the present application may be embodied in the form of a software product, where the computer software product is stored in a storage medium and includes one or more instructions to enable a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method of the embodiment of the present application. And the aforementioned storage medium may be a non-transitory storage medium comprising: a U-disk, a removable hard disk, a Read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other media capable of storing program codes, and may also be a transient storage medium.
The words used in this application are words of description only and not of limitation of the claims. As used in the description of the embodiments and the claims, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. Similarly, the term "and/or" as used in this application is meant to encompass any and all possible combinations of one or more of the associated listed. Furthermore, the terms "comprises" and/or "comprising," when used in this application, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
The various aspects, implementations, or features of the described embodiments can be used alone or in any combination. Aspects of the described embodiments may be implemented by software, hardware, or a combination of software and hardware. The described embodiments may also be embodied by a computer-readable medium having computer-readable code stored thereon, the computer-readable code comprising instructions executable by at least one computing device. The computer readable medium can be associated with any data storage device that can store data which can be read by a computer system. Exemplary computer readable media can include read-only memory, random-access memory, CD-ROMs, HDDs, DVDs, magnetic tape, and optical data storage devices, among others. The computer readable medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.
The above description of the technology may refer to the accompanying drawings, which form a part hereof, and in which is shown by way of illustration embodiments in which the described embodiments may be practiced. These embodiments, while described in sufficient detail to enable those skilled in the art to practice them, are non-limiting; other embodiments may be utilized and changes may be made without departing from the scope of the described embodiments. For example, the order of operations described in a flowchart is non-limiting, and thus the order of two or more operations illustrated in and described in accordance with the flowchart may be altered in accordance with several embodiments. As another example, in several embodiments, one or more operations illustrated in and described with respect to the flowcharts are optional or may be eliminated. Additionally, certain steps or functions may be added to the disclosed embodiments, or two or more steps may be permuted in order. All such variations are considered to be encompassed by the disclosed embodiments and the claims.
Additionally, terminology is used in the foregoing description of the technology to provide a thorough understanding of the described embodiments. However, no unnecessary detail is required to implement the described embodiments. Accordingly, the foregoing description of the embodiments has been presented for purposes of illustration and description. The embodiments presented in the foregoing description and the examples disclosed in accordance with these embodiments are provided solely to add context and aid in the understanding of the described embodiments. The above description is not intended to be exhaustive or to limit the described embodiments to the precise form disclosed. Many modifications, alternative uses, and variations are possible in light of the above teaching. In some instances, well known process steps have not been described in detail in order to avoid unnecessarily obscuring the described embodiments.

Claims (10)

1. An open parking space vehicle distinguishing method based on image recognition comprises the following steps:
s1, acquiring a picture to be detected, capturing a vehicle through a camera in parking lot parking space management equipment installed on the basis of the road side in an open parking lot, and acquiring the picture to be detected;
s2, detecting the license plate of the picture to be detected by using a license plate detection algorithm, judging whether the license plate exists in the picture, outputting license plate information of the picture when the license plate exists, and inputting the picture into the step S3 for processing when the license plate does not exist;
and S3, judging whether the car exists or not in the picture, if the car exists, outputting the picture that the car exists, and if the car does not exist, outputting the picture that the car does not exist.
2. The method for distinguishing vehicles on an open parking space according to claim 1, wherein the step of detecting the license plate of the picture to be detected by using a license plate detection algorithm and judging whether the license plate exists in the picture further comprises the following steps:
s2.1, obtaining a license plate area in the image by using a license plate detection algorithm;
and S2.2, license plate classification, namely cutting off the detected license plate region, and performing secondary classification by adopting a convolution network model to distinguish whether the license plate region is a license plate.
3. The method according to claim 2, wherein the license plate detection algorithm is a fast-RCNN algorithm, and the specific process is as follows:
1) inputting the whole picture into a convolutional neural network CNN to obtain a convolutional feature map;
2) inputting the convolution characteristic into an RPN (region pro-social network) to obtain characteristic information of the candidate frame;
3) judging whether the feature information extracted from the candidate frame belongs to a specific class by using a classifier;
4) and further adjusting the position of the candidate frame belonging to a certain characteristic by using a regressor.
4. The method according to claim 2, wherein the convolutional network model is a Resnet model obtained by a residual error unit in a convolutional neural network.
5. The method for identifying an open parking space vehicle according to claim 1, wherein the step S3 uses a universal mobilenet network to classify the image into six categories of exposed vehicle body wheels, exposed chassis, wheels, no vehicle, passing vehicle in the scene, and exposure of picture.
6. An open parking space vehicle discrimination device based on image recognition includes:
the image acquisition module to be detected captures a vehicle through a camera in parking lot parking space management equipment installed on the basis of the road side in an open parking lot to acquire an image to be detected;
the license plate detection module is used for detecting the license plate of the picture to be detected by utilizing a license plate detection algorithm, judging whether the license plate exists in the picture or not, outputting license plate information of the picture when the license plate exists, and inputting the picture into the classification module for classifying whether the vehicle exists or not when the license plate does not exist;
and the vehicle-presence and vehicle-absence classification module is used for performing vehicle-presence and vehicle-absence classification judgment on the picture, outputting the picture with a vehicle if the picture has a vehicle, and outputting the picture without a vehicle if the picture does not have a vehicle.
7. The device for distinguishing vehicles according to claim 6, wherein the step of detecting the license plate of the image to be detected by using a license plate detection algorithm and judging whether the license plate exists in the image further comprises the steps of:
the license plate region acquisition sub-module acquires a license plate region in the image by using a license plate detection algorithm;
and the license plate classification submodule cuts the detected license plate area, performs secondary classification by adopting a convolution network model, and distinguishes whether the license plate area is a license plate or not.
8. The apparatus according to claim 7, wherein the license plate detection algorithm is a fast-RCNN algorithm, and the specific process is as follows:
1) inputting the whole picture into a convolutional neural network CNN to obtain a convolutional feature map;
2) inputting the convolution characteristics into an RPN (region pro-social network) to obtain characteristic information of a candidate frame;
3) judging whether the feature information extracted from the candidate frame belongs to a specific class by using a classifier;
4) for a candidate box belonging to a feature, its position is further adjusted with a regressor.
9. The open parking space vehicle discrimination apparatus according to claim 7, wherein the convolutional network model is a Resnet model obtained by a residual error unit in a convolutional neural network.
10. The open parking space vehicle distinguishing device of claim 6, wherein the vehicle-in-vehicle-out classification module uses a universal mobilenet network to classify the images into six categories of exposed vehicle body wheels, exposed chassis only, wheels only, no vehicle, passing vehicle in the scene, and image exposure.
CN202010331657.1A 2020-04-23 2020-04-23 Open type parking space vehicle distinguishing method and device based on image recognition Active CN111508269B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010331657.1A CN111508269B (en) 2020-04-23 2020-04-23 Open type parking space vehicle distinguishing method and device based on image recognition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010331657.1A CN111508269B (en) 2020-04-23 2020-04-23 Open type parking space vehicle distinguishing method and device based on image recognition

Publications (2)

Publication Number Publication Date
CN111508269A true CN111508269A (en) 2020-08-07
CN111508269B CN111508269B (en) 2021-05-18

Family

ID=71877977

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010331657.1A Active CN111508269B (en) 2020-04-23 2020-04-23 Open type parking space vehicle distinguishing method and device based on image recognition

Country Status (1)

Country Link
CN (1) CN111508269B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112183472A (en) * 2020-10-28 2021-01-05 西安交通大学 Method for detecting whether test field personnel wear work clothes or not based on improved RetinaNet
CN112950954A (en) * 2021-02-24 2021-06-11 电子科技大学 Intelligent parking license plate recognition method based on high-position camera
CN114170808A (en) * 2021-12-14 2022-03-11 深圳市奥肯特科技有限公司 Device and method for collecting vehicle number plate in low power consumption mode
CN114566063A (en) * 2022-01-24 2022-05-31 深圳市捷顺科技实业股份有限公司 Intelligent parking space guiding management method and device and storage medium
CN115019296A (en) * 2022-08-04 2022-09-06 之江实验室 Cascading-based license plate detection and identification method and device

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104809443A (en) * 2015-05-05 2015-07-29 上海交通大学 Convolutional neural network-based license plate detection method and system
CN106022232A (en) * 2016-05-12 2016-10-12 成都新舟锐视科技有限公司 License plate detection method based on deep learning
CN106651765A (en) * 2016-12-30 2017-05-10 深圳市唯特视科技有限公司 Method for automatically generating thumbnail by use of deep neutral network
CN106652551A (en) * 2016-12-16 2017-05-10 浙江宇视科技有限公司 Parking stall detection method and device
CN106864369A (en) * 2015-12-11 2017-06-20 华创车电技术中心股份有限公司 Parking auxiliary ring field image system
CN107591018A (en) * 2017-08-03 2018-01-16 广州德中科技有限公司 A kind of open section parking management method and system
CN107610248A (en) * 2017-08-22 2018-01-19 罗云贵 A kind of parking charge apparatus and method
CN108198430A (en) * 2017-12-29 2018-06-22 智慧互通科技有限公司 A kind of trackside fence parking management system based on video camera and sensor
CN109146967A (en) * 2018-07-09 2019-01-04 上海斐讯数据通信技术有限公司 The localization method and device of target object in image
CN109344825A (en) * 2018-09-14 2019-02-15 广州麦仑信息科技有限公司 A kind of licence plate recognition method based on convolutional neural networks
CN110111313A (en) * 2019-04-22 2019-08-09 腾讯科技（深圳）有限公司 Medical image detection method and relevant device based on deep learning
CN110164139A (en) * 2019-05-30 2019-08-23 浙江大学 A kind of side parking detection identifying system and method
US10628688B1 (en) * 2019-01-30 2020-04-21 Stadvision, Inc. Learning method and learning device, and testing method and testing device for detecting parking spaces by using point regression results and relationship between points to thereby provide an auto-parking system
CN111553992A (en) * 2020-04-23 2020-08-18 深圳智优停科技有限公司 Parking space management equipment management method and system

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104809443A (en) * 2015-05-05 2015-07-29 上海交通大学 Convolutional neural network-based license plate detection method and system
CN106864369A (en) * 2015-12-11 2017-06-20 华创车电技术中心股份有限公司 Parking auxiliary ring field image system
CN106022232A (en) * 2016-05-12 2016-10-12 成都新舟锐视科技有限公司 License plate detection method based on deep learning
CN106652551A (en) * 2016-12-16 2017-05-10 浙江宇视科技有限公司 Parking stall detection method and device
CN106651765A (en) * 2016-12-30 2017-05-10 深圳市唯特视科技有限公司 Method for automatically generating thumbnail by use of deep neutral network
CN107591018A (en) * 2017-08-03 2018-01-16 广州德中科技有限公司 A kind of open section parking management method and system
CN107610248A (en) * 2017-08-22 2018-01-19 罗云贵 A kind of parking charge apparatus and method
CN108198430A (en) * 2017-12-29 2018-06-22 智慧互通科技有限公司 A kind of trackside fence parking management system based on video camera and sensor
CN109146967A (en) * 2018-07-09 2019-01-04 上海斐讯数据通信技术有限公司 The localization method and device of target object in image
CN109344825A (en) * 2018-09-14 2019-02-15 广州麦仑信息科技有限公司 A kind of licence plate recognition method based on convolutional neural networks
US10628688B1 (en) * 2019-01-30 2020-04-21 Stadvision, Inc. Learning method and learning device, and testing method and testing device for detecting parking spaces by using point regression results and relationship between points to thereby provide an auto-parking system
CN110111313A (en) * 2019-04-22 2019-08-09 腾讯科技（深圳）有限公司 Medical image detection method and relevant device based on deep learning
CN110164139A (en) * 2019-05-30 2019-08-23 浙江大学 A kind of side parking detection identifying system and method
CN111553992A (en) * 2020-04-23 2020-08-18 深圳智优停科技有限公司 Parking space management equipment management method and system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郑志锋: "基于深度学习的智能停车场车位查询系统", 《计算机应用》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112183472A (en) * 2020-10-28 2021-01-05 西安交通大学 Method for detecting whether test field personnel wear work clothes or not based on improved RetinaNet
CN112950954A (en) * 2021-02-24 2021-06-11 电子科技大学 Intelligent parking license plate recognition method based on high-position camera
CN114170808A (en) * 2021-12-14 2022-03-11 深圳市奥肯特科技有限公司 Device and method for collecting vehicle number plate in low power consumption mode
CN114566063A (en) * 2022-01-24 2022-05-31 深圳市捷顺科技实业股份有限公司 Intelligent parking space guiding management method and device and storage medium
CN115019296A (en) * 2022-08-04 2022-09-06 之江实验室 Cascading-based license plate detection and identification method and device

Also Published As

Publication number Publication date
CN111508269B (en) 2021-05-18

Similar Documents

Publication Publication Date Title
CN111508269B (en) Open type parking space vehicle distinguishing method and device based on image recognition
Baran et al. A smart camera for the surveillance of vehicles in intelligent transportation systems
CN112232237B (en) Method, system, computer device and storage medium for monitoring vehicle flow
US20230153698A1 (en) Methods and systems for accurately recognizing vehicle license plates
CN110738857A (en) vehicle violation evidence obtaining method, device and equipment
CN111027535A (en) License plate recognition method and related equipment
CN105608464A (en) Face recognition mobile terminal solving system and solving method
CN114529873A (en) Target detection method and city violation event monitoring method applying same
CN112135038A (en) Vehicle tracking method and system based on traffic monitoring and storage medium thereof
CN113112813B (en) Illegal parking detection method and device
Ashtari et al. A new reliable approach for Persian license plate detection on colour images
CN113221760A (en) Expressway motorcycle detection method
Zhang et al. A front vehicle detection algorithm for intelligent vehicle based on improved gabor filter and SVM
CN114973156B (en) Night muck car detection method based on knowledge distillation
Mago et al. Optimized outdoor parking system for smart cities using advanced saliency detection method and hybrid features extraction model
CN115861916A (en) Abnormal parking behavior detection method and device, computer equipment and storage medium
Prajapati et al. A Review Paper on Automatic Number Plate Recognition using Machine Learning: An In-Depth Analysis of Machine Learning Techniques in Automatic Number Plate Recognition: Opportunities and Limitations
Balbin et al. Cloud based color coding scheme violator plate detection through character recognition using image processing
CN113553953A (en) Vehicle parabolic detection method and device, electronic device and readable storage medium
Sharma et al. Automated Vehicle speed Estimation and License Plate Detection for Smart Cities Development
KR20210035360A (en) License Plate Recognition Method and Apparatus for roads
JP2021152826A (en) Information processing device, subject classification method, and subject classification program
CN113269011B (en) Vehicle detection method, device, equipment and storage medium
Fuad et al. A Novel Traffic System for Detecting Lane-Based Rule Violation
CN113435419B (en) Illegal garbage discarding behavior detection method, device and application

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
GR01 Patent grant
GR01 Patent grant