CN111898497B - License plate detection method, system, device and readable storage medium - Google Patents

License plate detection method, system, device and readable storage medium Download PDF

Info

Publication number
CN111898497B
CN111898497B CN202010686266.1A CN202010686266A CN111898497B CN 111898497 B CN111898497 B CN 111898497B CN 202010686266 A CN202010686266 A CN 202010686266A CN 111898497 B CN111898497 B CN 111898497B
Authority
CN
China
Prior art keywords
image
license plate
detected
neural network
position information
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
Application number
CN202010686266.1A
Other languages
Chinese (zh)
Other versions
CN111898497A (en
Inventor
路通
谢会斌
李聪廷
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.)
Jinan Boguan Intelligent Technology Co Ltd
Original Assignee
Jinan Boguan Intelligent 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 Jinan Boguan Intelligent Technology Co Ltd filed Critical Jinan Boguan Intelligent Technology Co Ltd
Priority to CN202010686266.1A priority Critical patent/CN111898497B/en
Publication of CN111898497A publication Critical patent/CN111898497A/en
Application granted granted Critical
Publication of CN111898497B publication Critical patent/CN111898497B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • 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
    • G06N3/08—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
    • G06V10/40—Extraction of image or video features
    • 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)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (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)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Image Analysis (AREA)

Abstract

The application discloses a license plate detection method, which comprises the following steps: acquiring an image to be detected; inputting an image to be detected into a positioning neural network, wherein the positioning neural network is a neural network with a semi-characteristic network structure convolved by using a DW (discrete wavelet transform); calculating the position information of the license plate in the image to be detected through a positioning neural network; and detecting the license plate of the image to be detected according to the position information. According to the application, the image to be detected is input into the positioning neural network with the DW (discrete wavelet) convolution semi-characteristic network structure, so that the position information of the license plate in the image to be detected can be calculated through the positioning neural network, then the license plate detection is carried out on the image to be detected according to the position information, the efficiency and the accuracy of the license plate detection are greatly improved, and the requirement of real-time detection on video streams can be met. The application also provides a license plate detection system, license plate detection equipment and a readable storage medium, and the license plate detection system has the beneficial effects.

Description

License plate detection method, system, device and readable storage medium
Technical Field
The present application relates to the field of license plate detection, and in particular, to a method, a system, an apparatus, and a readable storage medium for license plate detection.
Background
The real-time license plate detection needs to meet the requirements of the detection rate and the real-time performance at the same time, plays an important role in an intelligent traffic system, and can be effectively combined and applied through advanced deep learning technologies, sensing technologies and traditional imaging technologies at present, so that the license plate detection is presented in a new mode in a standard intelligent traffic scene.
However, under the complex traffic scene, a plurality of challenges exist in license plate positioning, and the license plate positioning is influenced by various external influences such as scene light, and meanwhile, along with the continuous increase of novel license plate types such as new energy vehicles, the traditional image judgment method is difficult to make effective response to scene and light changes, but the existing license plate detection technology based on deep learning has the defects that the time consumption is high due to overweight of a basic network, and the detection effect is reduced due to the fact that vehicles and license plates are detected, and further the efficiency and accuracy of license plate detection are low.
Therefore, how to improve the efficiency and accuracy of license plate detection is a technical problem that needs to be solved by those skilled in the art.
Disclosure of Invention
The application aims to provide a license plate detection method, a license plate detection system, license plate detection equipment and a readable storage medium, which are used for improving the efficiency and the accuracy of license plate detection.
In order to solve the technical problems, the application provides a license plate detection method, which comprises the following steps:
Acquiring an image to be detected;
Inputting the image to be detected into a positioning neural network, wherein the positioning neural network is a neural network with a semi-characteristic network structure convolved by using a DW (discrete wavelet transform);
Calculating the position information of the license plate in the image to be detected through the positioning neural network;
And detecting license plates of the images to be detected according to the position information.
Optionally, calculating, by the positioning neural network, position information of the license plate in the image to be detected includes:
carrying out DW convolution feature extraction on the image to be detected to obtain corresponding features;
The features are mixed and fused in disorder and divided into a first feature group and a second feature group;
performing DW convolution feature extraction on the first feature set, and combining the obtained features with the second feature set to obtain a third feature set;
and calculating the position information of the license plate in the image to be detected according to the third feature group.
Optionally, calculating the position information of the license plate in the image to be detected according to the third feature set includes:
Scaling the feature graphs in the third feature group;
And determining the size of a preselection frame of a target detection algorithm, and calculating the position information of the license plate in the image to be detected according to the third feature set by utilizing the preselection frame.
Optionally, after calculating the position information of the license plate in the image to be detected through the positioning neural network, the method further includes:
Calculating the confidence coefficient of the position information;
And outputting prompt information for replacing the image to be detected when the confidence coefficient is smaller than a threshold value.
Optionally, after acquiring the image to be detected, before inputting the image to be detected into the positioning neural network, the method further includes:
And carrying out format conversion on the image to be detected.
Optionally, before inputting the image to be detected into the positioning neural network, the method further includes:
acquiring a training set, and preprocessing training images in the training set; the preprocessing comprises at least one of overturning preprocessing, clipping and twisting preprocessing, gaussian blur preprocessing and random acquisition block domain preprocessing;
and training the positioning neural network by using the preprocessed training set.
The application also provides a license plate detection system, which comprises:
The acquisition module is used for acquiring the image to be detected;
the input module is used for inputting the image to be detected into a positioning neural network, wherein the positioning neural network is a neural network with a semi-characteristic network structure convolved by using a DW;
the first calculation module is used for calculating the position information of the license plate in the image to be detected through the positioning neural network;
and the license plate detection module is used for carrying out license plate detection on the image to be detected according to the position information.
Optionally, the first computing module includes:
the first extraction submodule is used for carrying out DW convolution feature extraction on the image to be detected to obtain corresponding features;
The division module is used for carrying out disorder fusion on all the features and dividing the features into a first feature group and a second feature group;
The second extraction submodule is used for carrying out DW convolution feature extraction on the first feature set, and combining the obtained features with the second feature set to obtain a third feature set;
And the calculating sub-module is used for calculating the position information of the license plate in the image to be detected according to the third feature group.
The application also provides license plate detection equipment, which comprises:
A memory for storing a computer program;
a processor for performing the steps of the method of license plate detection as claimed in any one of the preceding claims when executing the computer program.
The application also provides a readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the method of license plate detection as described in any of the preceding claims.
The license plate detection method provided by the application comprises the following steps: acquiring an image to be detected; inputting an image to be detected into a positioning neural network, wherein the positioning neural network is a neural network with a semi-characteristic network structure convolved by using a DW (discrete wavelet transform); calculating the position information of the license plate in the image to be detected through a positioning neural network; and detecting the license plate of the image to be detected according to the position information.
According to the technical scheme provided by the application, the image to be detected is input into the positioning neural network with the DW convolution semi-characteristic network structure, so that the time consumption of the positioning neural network is low, the position information of the license plate in the image to be detected can be calculated through the positioning neural network quickly, then the license plate detection is carried out on the image to be detected according to the position information, the license plate detection efficiency and accuracy are greatly improved, and the requirement of real-time detection on video streams can be met. The application also provides a license plate detection system, a license plate detection device and a readable storage medium, which have the beneficial effects and are not repeated here.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present application, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a license plate detection method according to an embodiment of the present application;
FIG. 2 is a flowchart of an actual implementation of S103 in the license plate detection method provided in FIG. 1;
FIG. 3 is a schematic diagram of a positioning neural network using a DW convolved semi-feature network architecture according to an embodiment of the present application;
FIG. 4 is a schematic diagram of function curves of prelu and relu activation functions according to an embodiment of the present application;
FIG. 5 is a block diagram of a license plate detection system according to an embodiment of the present application;
fig. 6 is a block diagram of a license plate detection device according to an embodiment of the present application.
Detailed Description
The core of the application is to provide a license plate detection method, a license plate detection system, license plate detection equipment and a readable storage medium, which are used for improving the license plate detection efficiency and accuracy.
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Based on the fact that a plurality of challenges exist in license plate positioning under the current complex traffic scene, the license plate positioning is influenced by various external influences such as scene light, and meanwhile, with the continuous increase of novel license plate types such as new energy vehicles, the traditional image judgment method is difficult to make effective response to scene and light changes, the existing license plate detection technology based on deep learning is high in time consumption caused by overweight of a basic network, and meanwhile, detection effect is reduced due to detection of vehicles and license plates, and further, license plate detection efficiency and accuracy are low; the present application provides a license plate detection method for solving the above problems.
Referring to fig. 1, fig. 1 is a flowchart of a license plate detection method according to an embodiment of the present application.
The method specifically comprises the following steps:
s101: acquiring an image to be detected;
Optionally, to further improve the efficiency and accuracy of license plate detection, after the image to be detected is acquired, format conversion may be performed on the image to be detected before the image to be detected is input into the positioning neural network.
In a specific embodiment, the application sets the input of the network to 300×160 instead of 224×224 according to the characteristic that the input image is 1920×1080, so that not only can a good detection of a small license plate target be ensured, but also distortion of the input image can be prevented.
S102: inputting an image to be detected into a positioning neural network;
The locating neural network is a neural network with a semi-characteristic network structure by using DW convolution, and the DW convolution is different from the conventional convolution operation, one convolution kernel is responsible for one channel, and one channel is only convolved by one convolution kernel, so that the network structure can reduce time consumption, and further, the license plate detection efficiency can be improved.
Optionally, to improve the robustness of the positioning neural network, the following steps may be further performed before inputting the image to be detected into the positioning neural network:
Acquiring a training set, and preprocessing training images in the training set; the preprocessing comprises at least one of overturning preprocessing, clipping and twisting preprocessing, gaussian blur preprocessing and random acquisition block domain preprocessing;
And training the positioning neural network by utilizing the preprocessed training set.
Because of a plurality of challenges in license plate positioning under the current complex traffic scene, the license plate positioning is influenced by various external influences such as scene light and the like and is continuously increased along with the new license plate types such as new energy vehicles and the like, the data set comprises the license plates of the motor vehicles used on the current road and as many complex vehicle road scenes as possible, such as scenes such as night, long distance, exposure, large angle, pollution and the like, and the positioning neural network is trained by using various training materials under the condition of insufficient data set, so that the robustness of the positioning neural network can be effectively improved.
S103: calculating the position information of the license plate in the image to be detected through a positioning neural network;
after the position information of the license plate in the image to be detected is calculated through the positioning neural network, license plate detection can be carried out on the image to be detected according to the position information, and further the efficiency and the accuracy of license plate detection are improved.
Optionally, the calculating, by using the positioning neural network, the position information of the license plate in the image to be detected may be implemented by executing the steps shown in fig. 2, referring to fig. 2, fig. 2 is a flowchart of an actual expression of S103 in the method for detecting the license plate provided in fig. 1, which specifically includes the following steps:
S201: carrying out DW convolution feature extraction on the image to be detected to obtain corresponding features;
s202: the method comprises the steps of (1) scrambling and fusing all the features, and dividing the features into a first feature group and a second feature group;
s203: performing DW convolution feature extraction on the first feature set, and combining the obtained features with the second feature set to obtain a third feature set;
s204: and calculating the position information of the license plate in the image to be detected according to the third feature set.
In a specific embodiment, please refer to fig. 3, fig. 3 is a schematic diagram of a positioning neural network using a semi-characteristic network structure of DW convolution according to an embodiment of the present application, as shown in fig. 3, the present application uses DW convolution instead of conventional convolution, where the convolution has the advantage of saving more than seven times of the computation power of a common convolution, for example, the normal input channel number 16 outputs 32, the convolution kernel size is 3x3, 16x32x3x3 = 4608, and DW is calculated to 16x3x3+16x32x1 = 656;
Furthermore, the application adopts a unique half-characteristic convolution mode, and the realization principle of the mode is that half characteristic values of the layer are reserved and are directly input to the next layer without being convolved by the next layer, so that calculation is reduced, characteristic fusion is increased, and deep semantic fusion is realized into shallow characteristics. The remote license plate can be better detected.
Preferably, in order to reduce the loss of the features, shuffleChannel (a method capable of improving the robustness of data) can be used in the positioning neural network to increase the information exchange between the features, the channel recombination of the neural network is completed through shuffleChannel, and the features are evenly divided into each group, so that the information of other groups can be obtained when each group is convolved, the direct communication effect between the groups is achieved, and the incomplete feature caused by the information sealing of each group due to DW convolution is compensated. In one embodiment, the group may be set to 2, as too many groups may result in an increase in MAC (physical memory):
And b= hwc1c2
Wherein h, w are the height and width of the feature map, g is the number of input and output channels divided into several groups, and c is the number of channels.
From the above formula, it can be seen that when B is unchanged, the larger g is, the larger the MAC is, and experiments show that when the group is 2, the MAC of the model is the smallest and the accuracy is the highest.
Preferably, in order to reduce the loss of the features, an activation function prelu may be used in the shallow layer to replace the activation function relu in the traditional network model, so as to achieve the effect of compensating for part of the missing features, please refer to fig. 4, fig. 4 is a schematic diagram of function curves of two activation functions, namely prelu and relu, provided in the embodiment of the present application, as shown in fig. 4, the prelu activation function retains a part of weights smaller than zero, and when the activation function is applied to the shallow layer, many small target information can be obtained, so that the detection rate of the license plate is greatly improved and the false detection rate is reduced.
The following is a set of experimental data obtained in connection with the present invention using prelu as an activation function.
Activation function name Mode of use Time consuming Accuracy rate of False detection rate
Relu All use 32fps 97.86％ 0.56％
PRelu All use 32fps 98.91％ 0.64％
PRelu The five layers are used 32fps 99.31％ 0.37％
Experiments show that the use of prelu activation functions not only increases the accuracy but also reduces false detection.
Optionally, the calculating, according to the third feature set, the position information of the license plate in the image to be detected may specifically be:
Scaling the feature graphs in the third feature group;
and determining the size of a preselection frame of the target detection algorithm, and calculating the position information of the license plate in the image to be detected according to the third feature set by utilizing the preselection frame.
In the prior art, the feature map is subjected to five times of scaling treatment, and the time consumption of network calculation is increased by multiple times of scaling, so that the time consumption can be reduced by reducing the times of scaling the feature map, and the detection rate of license plates is increased;
optionally, the application can also make appropriate adjustment to the network itself according to the scene of license plate detection, namely, the calculation mode of the preselection frame can be led into the concept of license plate aspect ratio, so that the calculation of the neural network is more biased to license plate detection, and the application can be realized by executing the following steps:
The calculation formula of the preselection frame is as follows:
The width of the rectangular large frame is according to the formula Calculating;
the height of the rectangular large frame is calculated according to a formula aspecct _ratio_min_size;
The side length of the square large frame is according to the formula Calculating;
The side length of the square small frame is min_size;
according to license plate detection characteristics, the small square pre-selection frame and the rectangular pre-selection frame with the height being larger than the width can be removed, because the two detection frames are not suitable for license plate detection and time consumption is increased, and in addition, the ratio of the pre-selection frames can be set to be 2:1 or 3:1 according to the aspect ratio which accords with license plate characteristics;
The aspect_ratio is the aspect ratio of the generated pre-selected frame, the min_size may be calculated according to the formula (input image size 10/100) + (input image size ratio/100), the max_size may be calculated according to the formula (input image size 20/100) + (input image size (ratio+step)/100), and the aspect ratio of the pre-frame may be calculated according to the formula (license plate width×license plate height)/(input image width×height).
Optionally, on the basis of the foregoing embodiment, the license plate detection is performed on the image to be detected according to the location information in step S104, which may specifically be the license plate detection performed on the image to be detected according to the location information and the license plate shape characteristic.
The detection rate and time-consuming comparison of the native target detection algorithm (Single Shot MultiBox Detector, SSD) with the optimized target detection algorithm:
the Arm end is time consuming Detection rate of False detection rate
Native SSD 40+Ms/frame 97.61％ 1.02％
Optimizing SSD 25 Ms/frame 99.53％ 0.17％
Experiments show that the SSD detection frame after optimization of the invention is reduced in time consumption and false detection rate, and the detection rate is improved to a certain extent.
Optionally, after calculating the position information of the license plate in the image to be detected through the positioning neural network, the following steps may be further performed:
calculating the confidence coefficient of the position information;
and outputting prompt information for replacing the image to be detected when the confidence coefficient is smaller than the threshold value.
In a specific embodiment, when the confidence coefficient is smaller than the threshold value, the detected position information is indicated to have low possibility of being license plate position information, at the moment, the prompt information of the image to be detected can be replaced, the area to be detected is replaced by a vehicle head area, the vehicle head area is acquired by a vehicle head detection network, and license plate detection is performed in the vehicle head area again, so that the possibility of detecting the license plate is improved.
S104: and detecting the license plate of the image to be detected according to the position information.
Based on the technical scheme, the method for detecting the license plate, provided by the application, has the advantages that the image to be detected is input into the positioning neural network with the DW convolution half-feature network structure, the time consumption of the positioning neural network is low due to the half-feature network structure, so that the position information of the license plate in the image to be detected can be calculated through the positioning neural network quickly, then the license plate detection is carried out on the image to be detected according to the position information, the efficiency and the accuracy of the license plate detection are greatly improved, and the requirement for real-time detection of video streams can be met.
Referring to fig. 5, fig. 5 is a block diagram of a license plate detection system according to an embodiment of the present application.
The system may include:
An acquisition module 100, configured to acquire an image to be detected;
The input module 200 is used for inputting the image to be detected into a positioning neural network, wherein the positioning neural network is a neural network with a semi-characteristic network structure convolved by using the DW;
the first calculating module 300 is used for calculating the position information of the license plate in the image to be detected through the positioning neural network;
the license plate detection module 400 is configured to detect a license plate of an image to be detected according to the position information.
Based on the above embodiments, in a preferred embodiment, the first computing module 300 may include:
The first extraction submodule is used for carrying out DW convolution feature extraction on the image to be detected to obtain corresponding features;
The division module is used for carrying out disorder fusion on all the features and dividing the features into a first feature group and a second feature group;
The second extraction submodule is used for carrying out DW convolution feature extraction on the first feature set, and combining the obtained features with the second feature set to obtain a third feature set;
And the calculating sub-module is used for calculating the position information of the license plate in the image to be detected according to the third feature group.
Further, the calculation sub-module may include:
a scaling processing unit, configured to perform scaling processing on the feature map in the third feature group;
And the calculating unit is used for determining the size of a pre-selection frame of the target detection algorithm and calculating the position information of the license plate in the image to be detected according to the third characteristic group by utilizing the pre-selection frame.
On the basis of the above embodiment, in a preferred embodiment, the system may further include:
The second calculation module is used for calculating the confidence coefficient of the position information;
And the output module is used for outputting prompt information for replacing the image to be detected when the confidence coefficient is smaller than the threshold value.
On the basis of the above embodiment, in a preferred embodiment, the system may further include:
The conversion module is used for carrying out format conversion on the image to be detected.
On the basis of the above embodiment, in a preferred embodiment, the system may further include:
The preprocessing module is used for acquiring a training set and preprocessing training images in the training set; the preprocessing comprises at least one of overturning preprocessing, clipping and twisting preprocessing, gaussian blur preprocessing and random acquisition block domain preprocessing;
And the training module is used for training the positioning neural network by utilizing the preprocessed training set.
Since the embodiments of the system portion and the embodiments of the method portion correspond to each other, the embodiments of the system portion refer to the description of the embodiments of the method portion, which is not repeated herein.
Referring to fig. 6, fig. 6 is a block diagram of a license plate detection apparatus according to an embodiment of the present application.
The license plate detection device 600 may vary considerably in configuration or performance and may include one or more processors (central processing units, CPU) 622 (e.g., one or more processors) and memory 632, one or more storage mediums 630 (e.g., one or more mass storage devices) that store applications 642 or data 644. Wherein memory 632 and storage medium 630 may be transitory or persistent storage. The program stored on the storage medium 630 may include one or more modules (not shown), each of which may include a series of instruction operations in the device. Still further, the processor 622 may be configured to communicate with the storage medium 630 and execute a series of instruction operations in the storage medium 630 on the license plate detection device 600.
License plate detection device 600 may also include one or more power supplies 626, one or more wired or wireless network interfaces 650, one or more input/output interfaces 658, and/or one or more operating systems 641, such as Windows ServerTM, mac OS XTM, unixTM, linuxTM, freeBSDTM, and the like.
The steps in the license plate detection method described above with reference to fig. 1 to 4 are implemented by the license plate detection apparatus based on the structure shown in fig. 6.
It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the systems, apparatuses and modules described above may refer to the corresponding processes in the foregoing method embodiments, which are not repeated herein.
In the several embodiments provided by the present application, it should be understood that the disclosed apparatus, device and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of modules is merely a logical function division, and there may be additional divisions of actual implementation, e.g., multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or modules, which may be in electrical, mechanical, or other forms.
The modules illustrated as separate components may or may not be physically separate, and components shown as modules may or may not be physical modules, i.e., may be located in one place, or may be distributed over a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist alone physically, or two or more modules may be integrated into one module. The integrated modules may be implemented in hardware or in software functional modules.
The integrated modules, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in whole or in part in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a function calling device, or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The method, the system, the equipment and the readable storage medium for license plate detection provided by the application are described in detail. The principles and embodiments of the present application have been described herein with reference to specific examples, the description of which is intended only to facilitate an understanding of the method of the present application and its core ideas. It should be noted that it will be apparent to those skilled in the art that various modifications and adaptations of the application can be made without departing from the principles of the application and these modifications and adaptations are intended to be within the scope of the application as defined in the following claims.
It should also be noted that in this specification, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises an element.

Claims (6)

1. A method for license plate detection, comprising:
Acquiring an image to be detected;
Inputting the image to be detected into a positioning neural network, wherein the positioning neural network is a neural network with a semi-characteristic network structure convolved by using a DW (discrete wavelet transform);
Calculating the position information of the license plate in the image to be detected through the positioning neural network;
carrying out license plate detection on the image to be detected according to the position information;
the calculating, by the positioning neural network, the position information of the license plate in the image to be detected includes:
Carrying out DW convolution feature extraction on the image to be detected to obtain corresponding features; the features are mixed and fused in disorder and divided into a first feature group and a second feature group; performing DW convolution feature extraction on the first feature set, and combining the obtained features with the second feature set to obtain a third feature set; calculating the position information of the license plate in the image to be detected according to the third feature set;
calculating the position information of the license plate in the image to be detected according to the third feature group, wherein the position information comprises:
Scaling the feature graphs in the third feature group; determining the size of a pre-selection frame of a target detection algorithm based on the aspect ratio of the license plate, and calculating the position information of the license plate in the image to be detected according to the third feature group by utilizing the pre-selection frame;
After calculating the position information of the license plate in the image to be detected through the positioning neural network, the method further comprises the following steps:
Calculating the confidence coefficient of the position information; and outputting prompt information for replacing the image to be detected when the confidence coefficient is smaller than a threshold value, so as to determine the headstock image as the image to be detected after replacement.
2. The method of claim 1, wherein after acquiring the image to be detected, before inputting the image to be detected into a localization neural network, further comprising:
And carrying out format conversion on the image to be detected.
3. The method of claim 1, further comprising, prior to inputting the image to be detected into a localization neural network:
acquiring a training set, and preprocessing training images in the training set; the preprocessing comprises at least one of overturning preprocessing, clipping and twisting preprocessing, gaussian blur preprocessing and random acquisition block domain preprocessing;
and training the positioning neural network by using the preprocessed training set.
4. A system for license plate detection, comprising:
The acquisition module is used for acquiring the image to be detected;
the input module is used for inputting the image to be detected into a positioning neural network, wherein the positioning neural network is a neural network with a semi-characteristic network structure convolved by using a DW;
the first calculation module is used for calculating the position information of the license plate in the image to be detected through the positioning neural network;
the license plate detection module is used for carrying out license plate detection on the image to be detected according to the position information;
wherein the first computing module comprises:
the first extraction submodule is used for carrying out DW convolution feature extraction on the image to be detected to obtain corresponding features;
The division module is used for carrying out disorder fusion on all the features and dividing the features into a first feature group and a second feature group;
The second extraction submodule is used for carrying out DW convolution feature extraction on the first feature set, and combining the obtained features with the second feature set to obtain a third feature set;
The calculating sub-module is used for calculating the position information of the license plate in the image to be detected according to the third feature group;
the calculating submodule is specifically configured to:
Scaling the feature graphs in the third feature group; determining the size of a pre-selection frame of a target detection algorithm based on the aspect ratio of the license plate, and calculating the position information of the license plate in the image to be detected according to the third feature group by utilizing the pre-selection frame;
the license plate detection system is also specifically used for:
Calculating the confidence coefficient of the position information; and outputting prompt information for replacing the image to be detected when the confidence coefficient is smaller than a threshold value, so as to determine the headstock image as the image to be detected after replacement.
5. A license plate detection apparatus, characterized by comprising:
A memory for storing a computer program;
a processor for implementing the steps of the method of license plate detection of any one of claims 1 to 3 when executing the computer program.
6. A readable storage medium, characterized in that it has stored thereon a computer program which, when executed by a processor, implements the steps of the method of license plate detection according to any of claims 1 to 3.
CN202010686266.1A 2020-07-16 2020-07-16 License plate detection method, system, device and readable storage medium Active CN111898497B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010686266.1A CN111898497B (en) 2020-07-16 2020-07-16 License plate detection method, system, device and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010686266.1A CN111898497B (en) 2020-07-16 2020-07-16 License plate detection method, system, device and readable storage medium

Publications (2)

Publication Number Publication Date
CN111898497A CN111898497A (en) 2020-11-06
CN111898497B true CN111898497B (en) 2024-05-10

Family

ID=73189205

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010686266.1A Active CN111898497B (en) 2020-07-16 2020-07-16 License plate detection method, system, device and readable storage medium

Country Status (1)

Country Link
CN (1) CN111898497B (en)

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105405298A (en) * 2015-12-24 2016-03-16 浙江宇视科技有限公司 Identification method of license plate identifications and device
EP3182334A1 (en) * 2015-12-17 2017-06-21 Xerox Corporation License plate recognition using coarse-to-fine cascade adaptations of convolutional neural networks
CN109447069A (en) * 2018-10-31 2019-03-08 沈阳工业大学 Collecting vehicle information recognition methods and system towards intelligent terminal
CN109840521A (en) * 2018-12-28 2019-06-04 安徽清新互联信息科技有限公司 A kind of integrated licence plate recognition method based on deep learning
CN110533103A (en) * 2019-08-30 2019-12-03 的卢技术有限公司 A kind of lightweight wisp object detection method and system
WO2020038205A1 (en) * 2018-08-24 2020-02-27 腾讯科技（深圳）有限公司 Target detection method and apparatus, computer-readable storage medium, and computer device
CN111209858A (en) * 2020-01-06 2020-05-29 电子科技大学 Real-time license plate detection method based on deep convolutional neural network
CN111401361A (en) * 2020-03-06 2020-07-10 南京理工大学 End-to-end lightweight deep license plate recognition method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3182334A1 (en) * 2015-12-17 2017-06-21 Xerox Corporation License plate recognition using coarse-to-fine cascade adaptations of convolutional neural networks
CN105405298A (en) * 2015-12-24 2016-03-16 浙江宇视科技有限公司 Identification method of license plate identifications and device
WO2020038205A1 (en) * 2018-08-24 2020-02-27 腾讯科技（深圳）有限公司 Target detection method and apparatus, computer-readable storage medium, and computer device
CN109447069A (en) * 2018-10-31 2019-03-08 沈阳工业大学 Collecting vehicle information recognition methods and system towards intelligent terminal
CN109840521A (en) * 2018-12-28 2019-06-04 安徽清新互联信息科技有限公司 A kind of integrated licence plate recognition method based on deep learning
CN110533103A (en) * 2019-08-30 2019-12-03 的卢技术有限公司 A kind of lightweight wisp object detection method and system
CN111209858A (en) * 2020-01-06 2020-05-29 电子科技大学 Real-time license plate detection method based on deep convolutional neural network
CN111401361A (en) * 2020-03-06 2020-07-10 南京理工大学 End-to-end lightweight deep license plate recognition method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
使用多任务级联卷积神经网络进行车牌照识别;胡从坤;黄东军;;企业技术开发(第02期);正文第1-5节 *
基于深度学习的实时车牌检测;李兴维;朱泽群;万秋波;屈金山;;信息通信(第12期);全文 *

Also Published As

Publication number Publication date
CN111898497A (en) 2020-11-06

Similar Documents

Publication Publication Date Title
CN111104903B (en) Depth perception traffic scene multi-target detection method and system
CN111242127A (en) Vehicle detection method with granularity level multi-scale characteristics based on asymmetric convolution
Xiang et al. Lightweight fully convolutional network for license plate detection
CN114419413A (en) Method for constructing sensing field self-adaptive transformer substation insulator defect detection neural network
CN111612024A (en) Feature extraction method and device, electronic equipment and computer-readable storage medium
CN116580195A (en) Remote sensing image semantic segmentation method and system based on ConvNeXt convolution
CN114529583B (en) Power equipment tracking method and tracking system based on residual regression network
CN110503049B (en) Satellite video vehicle number estimation method based on generation countermeasure network
CN111898497B (en) License plate detection method, system, device and readable storage medium
US20110222759A1 (en) Information processing apparatus, information processing method, and program
CN114862925A (en) Image registration method, device and system based on SIFT and storage medium
CN102013101A (en) Blind detection method of permuted and tampered images subjected to fuzzy postprocessing
CN114202473A (en) Image restoration method and device based on multi-scale features and attention mechanism
CN112101113B (en) Lightweight unmanned aerial vehicle image small target detection method
CN111444913B (en) License plate real-time detection method based on edge guiding sparse attention mechanism
CN106023184A (en) Depth significance detection method based on anisotropy center-surround difference
CN117132910A (en) Vehicle detection method and device for unmanned aerial vehicle and storage medium
CN107820087A (en) According to the method for mobile detection result dynamic regulation code check
CN112381088A (en) License plate recognition method and system for oil tank truck
CN113496159B (en) Multi-scale convolution and dynamic weight cost function smoke target segmentation method
Mei et al. Study of the eye-tracking methods based on video
CN113780136A (en) VOCs gas leakage detection method, system and equipment based on space-time texture recognition
CN115346143A (en) Behavior detection method, electronic device, and computer-readable medium
CN112818836A (en) Personnel target detection method and system for transformer substation scene
CN109919979A (en) A kind of method of video real-time modeling 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
GR01 Patent grant
GR01 Patent grant