CN110414507B - License plate recognition method and device, computer equipment and storage medium - Google Patents

License plate recognition method and device, computer equipment and storage medium Download PDF

Info

Publication number
CN110414507B
CN110414507B CN201910626156.3A CN201910626156A CN110414507B CN 110414507 B CN110414507 B CN 110414507B CN 201910626156 A CN201910626156 A CN 201910626156A CN 110414507 B CN110414507 B CN 110414507B
Authority
CN
China
Prior art keywords
license plate
image
detection model
candidate
region
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
CN201910626156.3A
Other languages
Chinese (zh)
Other versions
CN110414507A (en
Inventor
陈盈全
鲁继勇
洪国恩
赖胜军
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 CN201910626156.3A priority Critical patent/CN110414507B/en
Publication of CN110414507A publication Critical patent/CN110414507A/en
Application granted granted Critical
Publication of CN110414507B publication Critical patent/CN110414507B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/63—Scene text, e.g. street names
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/625—License plates

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)
  • Character Input (AREA)

Abstract

The embodiment of the invention discloses a license plate identification method, which comprises the following steps: detecting an image containing a license plate to obtain a detected candidate license plate region; taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model; correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image; and identifying license plate characters in the target license plate image to obtain an identification result. The license plate recognition method can accurately recognize the shot large-angle license plate, and greatly improves the accuracy of the large-angle license plate recognition. In addition, a license plate recognition apparatus, a computer device and a storage medium are also provided.

Description

License plate recognition method and device, computer equipment and storage medium
Technical Field
The invention relates to the technical field of license plate recognition, in particular to a license plate recognition method and device, computer equipment and a storage medium.
Background
At present, license plate recognition is mostly based on scenes such as a standard checkpoint and a parking lot, and in the captured pictures of the scenes, general license plates are normal, have no inclination and are relatively clear, and the accuracy of the existing license plate recognition algorithm is high for the general scenes.
At present, with the increase of the number of automobiles in China, roadside parking is urgently required to be standardized, and license plate judgment, image duplication elimination and license plate recognition are required to be carried out according to roadside parking scenes. Because the license plate angle is large (namely the inclination is large) in the license plate picture shot by the front-end camera of the roadside scene, the license plate is easy to have large distortion and is easy to be influenced by illumination, the recognition rate of the current license plate recognition algorithm for the scene is low.
Disclosure of Invention
Therefore, in order to solve the above problems, a license plate recognition method, a license plate recognition device, a computer device, and a storage medium are needed to accurately recognize a license plate with a large angle and distortion.
A license plate recognition method is characterized by comprising the following steps:
detecting an image containing a license plate to obtain a detected candidate license plate area;
taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model;
correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image;
and identifying characters in the target license plate image to obtain an identification result.
A license plate recognition device, the device comprising:
the acquisition module is used for detecting an image containing a license plate and acquiring a detected candidate license plate area;
the detection module is used for taking the candidate license plate area as the input of a license plate vertex detection model and acquiring the positions of four license plate vertexes obtained by detection output by the license plate vertex detection model;
the correction module is used for correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image;
and the recognition module is used for recognizing the characters in the target license plate image to obtain a recognition result.
A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of:
detecting an image containing a license plate to obtain a detected candidate license plate region;
taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model;
correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image;
and identifying characters in the target license plate image to obtain an identification result.
A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
detecting an image containing a license plate to obtain a detected candidate license plate area;
taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertexes obtained by detection output by the license plate vertex detection model;
correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image;
and identifying license plate characters in the target license plate image to obtain an identification result.
According to the license plate recognition method, the license plate recognition device, the computer equipment and the storage medium, the detected candidate license plate region is obtained by detecting the image containing the license plate, then the candidate license plate region is used as the input of the license plate vertex detection model, the positions of four license plate vertexes obtained by detection output by the license plate vertex detection model are obtained, then the license plate image is corrected according to the positions of the four license plate vertexes to obtain the corrected target license plate image, and then the license plate characters in the target license plate image are recognized to obtain the recognition result. The license plate identification method can accurately identify the shot large-angle license plate and the distorted license plate, and greatly improves the accuracy of identification of the large-angle license plate and the distorted license plate.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Wherein:
FIG. 1 is a flow diagram of a license plate recognition method in one embodiment;
FIG. 2 is a schematic diagram illustrating an exemplary effect of identifying a large-angle license plate image;
FIG. 3 is a flowchart of a method for license plate vertex detection in one embodiment;
FIG. 4 is a flow diagram of a method for detecting candidate license plate regions in accordance with one embodiment;
FIG. 5 is a schematic diagram of a license plate recognition process according to an embodiment;
FIG. 6 is a block diagram of a license plate recognition device according to an embodiment;
FIG. 7 is a block diagram of the structure of a detection module in one embodiment;
FIG. 8 is a block diagram of the architecture of an acquisition module in one embodiment;
FIG. 9 is a block diagram that illustrates the architecture of a computing device in one embodiment.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, in an embodiment, a license plate recognition method is provided, which may be applied to a terminal or a server, and specifically includes the following steps:
step 102, detecting the image containing the license plate, and acquiring a detected candidate license plate area.
The candidate license plate area refers to an area where a detected license plate is located. There are many ways to detect an image containing a license plate to obtain a candidate license plate region, for example, an eight-connected region search algorithm may be used to search for a license plate region.
And step 104, taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model.
The area of the candidate license plate region is far larger than that of the license plate image, so that the positions of four license plate vertexes are required to be detected in order to accurately position the positions of the license plate, and the license plate vertexes refer to points of four corners of the license plate. The license plate vertex detection model is a model which is obtained through training and is used for detecting four license plate vertexes in a candidate license plate area. In one embodiment, the license plate vertex detection model can be obtained by training a convolutional neural network model. The training of the license plate vertex detection model can be obtained by adopting supervised training, a training sample and the label of the sample are obtained, the training sample refers to a candidate license plate area sample, and the label of the sample refers to the position coordinates of the four corners of the license plate labeled in the candidate license plate area sample.
And step 106, correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image.
Since the license plate image may be an image at a large angle or a distorted image (for example, an image obtained by shooting at a low angle), after the position coordinates of the four license plate vertexes are located, the license plate image needs to be corrected according to the positions of the four license plate vertexes, that is, the license plate is converted into a relatively correct license plate image, so that subsequent identification is facilitated.
Fig. 2 is a schematic diagram illustrating an effect of recognizing a large-angle license plate image in one embodiment. It can be seen from the figure that the photographed license plate image with the deviation is corrected to obtain a more correct license plate image.
And step 108, identifying license plate characters in the target license plate image to obtain an identification result.
The target license plate image is a corrected and relatively positive license plate image, the license plate image comprises license plate characters, and the license plate characters comprise letters and numbers. And in order to obtain the license plate number, identifying the target license plate image to obtain an identification result. Because the corrected target license plate image is the same as the license plate image obtained by normal shooting, the license plate number can be identified by adopting a traditional identification algorithm. For example, a method based on connected region search and vertical projection can be adopted to segment license plate characters, and then the segmented license plate characters are identified by a character identification model obtained through deep learning training to obtain an identification result of each character, so that a license plate number obtained through identification is obtained.
According to the license plate recognition method, the image containing the license plate is detected, the detected candidate license plate area is obtained, then the candidate license plate area is used as the input of the license plate vertex detection model, the positions of four license plate vertexes detected and output by the license plate vertex detection model are obtained, then the license plate image is corrected according to the positions of the four license plate vertexes, the corrected target license plate image is obtained, and then license plate characters in the target license plate image are recognized, so that the recognition result is obtained. The license plate identification method can accurately identify the shot large-angle license plate and the distorted license plate, and greatly improves the accuracy of identification of the large-angle license plate and the distorted license plate.
As shown in FIG. 3, in one embodiment, the license plate vertex detection model includes: a first detection model, a second detection model and a third detection model; taking the candidate license plate area as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertexes obtained by the detection of the output of the license plate vertex detection model, wherein the steps comprise:
and step 104A, taking the candidate license plate region as an input of a first detection model, wherein the first detection model is used for detecting the candidate license plate region to generate a candidate license plate window.
The license plate vertex detection model comprises three submodels, namely a first detection model, a second detection model and a third detection model. The first detection model is used to generate a candidate window. The candidate windows may be selected for windows having the same license plate size.
In one embodiment, the first inspection model is a model composed of convolutional layers, including: three convolutional layers, with an input image width of 16 and a height of 12, have convolution kernels of 3X3 for each convolutional layer.
And step 104B, obtaining a candidate license plate window output by the first detection model, and correcting the candidate license plate window to obtain a first candidate license plate window.
The correction processing refers to correcting the position of the candidate license plate window. After the candidate license plate window output by the first detection model is obtained, the generated candidate license plate window is not accurate, so that the candidate license plate window needs to be corrected to obtain a first candidate license plate window. In another embodiment, an NMS (Non-Maximum compression) can be used for combining candidate license plate windows with high overlapping rate after the correction processing.
And step 104C, taking the first candidate license plate window as the input of a second detection model, wherein the second detection model is used for screening the first candidate license plate window.
The second detection model screens the obtained plurality of first candidate license plate windows, and rejects a large number of existing non-license plate windows, so that more accurate candidate license plate windows are obtained.
In one embodiment, the second inspection model is also composed of convolution layers, the second inspection model inputs an image with a width of 36 and a height of 24, and comprises: three convolutional layers and a full connection layer, wherein 3X3 is selected as the convolutional core of the first convolutional layer and the convolutional core of the second convolutional layer, and 2X2 is selected as the convolutional core of the third convolutional layer.
And step 104D, correcting the candidate license plate window output by the second detection model to obtain a second candidate license plate window.
And similarly, correcting the candidate license plate window output by the second detection model to obtain a second candidate license plate window. The correction mode can adopt a license plate frame regression vector to correct the candidate license plate window. And then, combining the candidate license plate windows with high overlapping rate by adopting NMS (network management system) to obtain a second candidate license plate window.
And step 104E, taking the second candidate license plate window as the input of a third detection model, wherein the third detection model is used for identifying license plate areas to obtain the positions of four license plate vertexes.
The third detection model is a model obtained by adopting supervision training and is used for identifying the license plate area and then determining the positions of four license plate vertexes. In order to improve the recognition accuracy, the third detection model needs to be trained by adopting more training samples during training, and the samples of the third detection model are marked as the position coordinates of the four license plate vertexes, so that the trained third detection model can directly obtain the positions of the four license plate vertexes.
In one embodiment, the third detection model comprises: four convolutional layers and 1 fully-connected layer, the convolution kernel of the first convolutional layer, the second convolutional layer and the third convolutional layer is 3X3, the convolution kernel of the fourth convolutional layer is 2X2, the width of the input image of the third detection model is 48, and the height is 72.
The first detection model, the second detection model and the third detection model are all realized by adopting a convolutional neural network model. The first detection model, the second detection model and the third detection model are used together to accurately position the four vertexes of the license plate, and the accuracy of subsequent recognition is improved.
In one embodiment, obtaining a candidate license plate window output by a first detection model, and performing correction processing on the candidate license plate window to obtain a first candidate license plate window includes: obtaining a candidate license plate window and a license plate frame regression vector output by the first detection model; and correcting the candidate license plate window by adopting the license plate frame regression vector to obtain a first candidate license plate window.
The first detection model outputs a candidate license plate window and also outputs a license plate frame regression vector, and the license plate frame regression vector is adopted to correct the candidate license plate window to obtain a first candidate license plate window.
In one embodiment, the correcting the license plate image according to the positions of four license plate vertexes to obtain a corrected target license plate image includes: calculating according to the positions of the four license plate vertexes to obtain a perspective transformation matrix; and correcting the license plate image according to the perspective transformation matrix to obtain a target license plate image.
The position coordinates (namely, source coordinates) of four license plate vertexes obtained through detection are known, and target position coordinates (namely, target coordinates) of the four license plate vertexes needing to be transformed are set, so that the perspective transformation matrix can be obtained through calculation according to the four source coordinates and the four target coordinates. And then, transforming the coordinates of each pixel in the license plate image according to the computed perspective transformation matrix to obtain a corrected target license plate image.
The formula for the perspective transformation can be expressed as follows:
Figure BDA0002127175220000071
(u, v, w) refers to the coordinates before transformation, (x ', y ', z ') refers to the coordinates after transformation, and w is 1 for a two-dimensional image. Wherein, x is x '/w', and y is y '/w'.
As shown in fig. 4, in an embodiment, detecting an image including a license plate to obtain a detected candidate license plate region includes:
step 102A, acquiring an original image containing a license plate, and preprocessing the original image, wherein the preprocessing comprises the following steps: and at least one of gray processing, scaling processing and denoising processing.
The process of detecting the image of the license plate can be placed on front-end equipment, namely a terminal, or on background equipment. The original image is an original image which is obtained by shooting and contains a license plate. The gray processing refers to converting an image into a gray image, so that the detection speed is increased. The scaling process refers to scaling the image to a preset size. The denoising process is to remove noise in an image. The denoising process may be performed by using gaussian filter convolution.
And step 102B, carrying out vertical edge detection on the preprocessed image, and determining the height and width of the image.
The vertical edge detection means detecting a vertical edge map of an image, and determining four edges of the image, thereby determining the height and width of the image. The vertical edge detection may be filtered using a filter.
And step 102C, dividing the image into a plurality of image blocks according to the height and the width of the image.
Before the image is binarized, the image is divided into a plurality of image blocks according to the height and the width of the image, so that the image features can be more accurately highlighted, and the subsequent binarization for each image block is facilitated. If the whole vehicle is subjected to binarization, a lot of information is easily lost, and subsequent license plate recognition is influenced.
In one embodiment, assuming that the image has a width W and a height H, the width and height are divided into m and n shares, respectively, for a total of m × n block regions.
And 102D, determining a binarization threshold corresponding to each image block according to the pixel values in the image blocks.
Before each image block is binarized, a binarization threshold corresponding to each image block is determined. In one embodiment, the pixel values of each pixel in the image block may be summed and then averaged, with the average being used as the binarization threshold for the image block. Pixels larger than the binarization threshold are set to 255, otherwise to 0.
In another embodiment, the maximum pixel value and the minimum pixel value in each image block are obtained respectively, and the maximum pixel value and the minimum pixel value are averaged to obtain a corresponding comparison value. The calculation of the alignment value can be represented by the following formula:
Thr i ＝(max+min)/2
wherein max is the maximum pixel value in the region, min is the minimum pixel value in the region, and then each point in the region is binarized by using the following formula:
Figure BDA0002127175220000091
wherein, B is the value of each pixel point of the binary image, v is the pixel value of the midpoint of the image, Thr i The comparison value of the block to which the pixel belongs, T is the threshold value of binarization, where 5 can be taken, or adaptive according to the sceneAnd (6) finishing. That is, the absolute value of the difference between the pixel value corresponding to each pixel point and the comparison value is calculated, and then the pixel point with the absolute value of the difference larger than the set binarization threshold is set to be 255, otherwise, the pixel point is set to be 0.
And step 102E, performing binarization processing on the image blocks according to the comparison value of each image block and a preset binarization threshold value to obtain a binarization image.
And carrying out binarization processing on the image blocks according to the comparison value of each image block and a preset binarization threshold value so as to obtain a binarization image.
And step 102F, detecting the license plate region in the binary image to obtain a candidate license plate region.
And detecting the license plate region aiming at the binary image to obtain a candidate license plate region obtained by detection. The detection of the license plate region may be performed in an existing manner, for example, a search manner of a connected region may be used for the detection, and the detection method is not limited herein.
The process of obtaining the candidate license plate region can more accurately obtain the candidate license plate region, and the accuracy of subsequent license plate recognition is improved conveniently.
In one embodiment, detecting a license plate region in a binarized image to obtain candidate license plate regions includes: carrying out corrosion expansion processing on each detected region by adopting a morphological algorithm on the binary image to obtain an image containing a communicated region; conducting connected region search on the images containing the connected regions to obtain suspected license plate regions; and detecting according to the texture characteristics of each suspected license plate area, and removing the pseudo license plate areas to obtain candidate license plate areas.
The fracture areas can be combined by adopting morphological algorithms such as corrosion expansion and the like, and meanwhile, the interference of small areas is eliminated, so that the image containing the connected areas is obtained. And then, performing connected region search on the processed image by adopting an eight-connected region search algorithm to search a suspected license plate region, then identifying by adopting texture characteristics to remove a fake license plate region, and finally obtaining a candidate license plate region.
The texture feature identification may specifically adopt the following manner: and carrying out binarization processing on the suspected license plate area, then counting the number of lines L with jump larger than a set threshold value in the area, and if the number of lines L is smaller than a preset threshold value T, removing the suspected license plate area, wherein the preset threshold value T can be half of the height of the area.
In one embodiment, recognizing characters in a target license plate image to obtain a recognition result, includes: detecting and segmenting characters in the target license plate image to obtain license plate characters;
and taking the license plate characters as the input of a character recognition model, and acquiring a recognition result output by the character recognition model, wherein the character recognition model is obtained by adopting deep learning algorithm training.
The character detection method can adopt a connected region search and vertical projection method, and the characters are segmented according to the detected characters to obtain individual characters, and then each character is identified. And pre-training to obtain a character recognition model, inputting the license plate characters obtained by segmentation as the character recognition model, and obtaining a recognition result output by the character recognition model.
In another embodiment, recognizing characters in a target license plate image to obtain a recognition result includes: detecting and segmenting characters in the target license plate image to obtain license plate character images; carrying out normalization processing on the license plate character image to obtain a target license plate character image; extracting the characteristics of the target license plate character image, and recognizing characters according to the extracted characteristics; and obtaining the confidence coefficient of character recognition, and when the confidence coefficient is smaller than a preset threshold value, taking the corresponding target license plate character image as the input of a character recognition model, and obtaining the recognition result output by the character recognition model, wherein the character recognition model is obtained by adopting deep learning algorithm training. Carrying out normalization processing on the license plate character image to obtain a target license plate character image; extracting the characteristics of the target license plate character image, and recognizing the characters according to the extracted characteristics; and obtaining the confidence coefficient of character recognition, and when the confidence coefficient is smaller than a preset threshold value, taking the corresponding target license plate character image as the input of a character recognition model, and obtaining the recognition result output by the character recognition model, wherein the character recognition model is obtained by adopting deep learning algorithm training.
The characters which can be segmented are unified to the size of 24 pixels with the width of 48 and the height, a Gabor feature of the character image is extracted (the Gabor feature is a feature which can be used for describing image texture information), an SVM (Support Vector Machine, which is a common discrimination method) is adopted for character recognition, when the confidence coefficient of the character recognition is smaller than a set threshold value, a convolutional neural network (namely a character recognition model) is adopted for carrying out secondary recognition on the characters, and then the result is output. Experiments prove that the accuracy rate of the Gabor feature extraction combined with the SVM method for recognizing clear characters is high, and the convolutional neural network has good effects of recognizing dirty characters with low contrast, so that the accuracy rate of the overall recognition of the characters is higher than 99.9% by combining the two methods.
As shown in fig. 5, in one embodiment, a schematic diagram of a license plate recognition process is provided. The method comprises the following steps: the method comprises three stages, namely a first stage of license plate detection, a second stage of license plate correction, a third stage of license plate identification.
The license plate detection in the first stage comprises the following steps: image preprocessing, vertical edge detection, block binaryzation, morphological processing, license plate area searching, fake license plate area removing and the like.
The second stage of license plate correction comprises the following steps: and identifying four license plate vertexes and correcting the license plate.
The license plate recognition of the third stage comprises the following steps: character segmentation and character recognition.
In one embodiment, the license plate detection can be completed by the front-end device, and then the license plate correction and the license plate recognition can be realized by the cloud (i.e. the server). And in the first stage, if the license plate is not detected to be contained, the license plate is not uploaded to the cloud.
As shown in fig. 6, in one embodiment, a license plate recognition device is provided, including:
an obtaining module 602, configured to detect an image including a license plate, and obtain a detected candidate license plate region;
the detection module 604 is configured to use the candidate license plate region as an input of a license plate vertex detection model, and obtain positions of four license plate vertices detected by the output of the license plate vertex detection model;
the correcting module 606 is used for correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image;
the recognition module 608 is configured to recognize license plate characters in the target license plate image to obtain a recognition result.
As shown in fig. 7, in one embodiment, the license plate vertex detection model includes: a first detection model, a second detection model and a third detection model; the detection module 604 includes:
a first detection module 604A, configured to take the candidate license plate region as an input of the first detection model, where the first detection model is configured to detect the candidate license plate region and generate a candidate license plate window; acquiring a candidate license plate window output by the first detection model, and correcting the candidate license plate window to obtain a first candidate license plate window;
a second detection module 604B, configured to use the first candidate license plate window as an input of the second detection model, where the second detection model is used to filter the first candidate license plate window; correcting the candidate license plate window output by the second detection model to obtain a second candidate license plate window;
the third detection module 604C is configured to use the second candidate license plate window as an input of the third detection model, where the third detection model is configured to identify a license plate region, and obtain positions of four license plate vertices.
In one embodiment, the first detection module 604A is further configured to obtain a license plate window candidate and a license plate frame regression vector output by the first detection module; and correcting the license plate window by adopting the license plate frame regression vector to obtain the first candidate license plate window.
In one embodiment, the correction module is further configured to calculate a perspective transformation matrix according to the positions of the four license plate vertices; and correcting the license plate image according to the perspective transformation matrix to obtain a target license plate image.
As shown in fig. 8, in one embodiment, the obtaining module 602 includes:
the preprocessing module 602A is configured to obtain an original image including a license plate, and perform preprocessing on the original image, where the preprocessing includes: at least one of gray processing, scaling processing and denoising processing;
an edge detection module 602B, configured to perform vertical edge detection on the preprocessed image, and determine the height and width of the image;
a block binarization module 602C, configured to divide the image into a plurality of image blocks according to the height and the width of the image; determining a comparison value corresponding to each image block according to pixel values in the image blocks; carrying out binarization processing on the image blocks according to the comparison value of each image block and a preset binarization threshold value to obtain a binarization image;
and the region detection module 602D is configured to detect a license plate region in the binarized image to obtain a candidate license plate region.
In an embodiment, the region detection module 602D is further configured to perform erosion and expansion processing on each detected region by using a morphological algorithm on the binarized image to obtain an image including a connected region; conducting connected region search on the images containing the connected regions to obtain suspected license plate regions; and detecting according to the texture characteristics of each suspected license plate area, and removing the pseudo license plate areas to obtain candidate license plate areas.
In one embodiment, the recognition module is further configured to detect and segment characters in the target license plate image to obtain individual license plate character images; carrying out normalization processing on the license plate character image to obtain a target license plate character image; extracting the characteristics of the target license plate character image, and recognizing characters according to the extracted characteristics; and acquiring a confidence coefficient of character recognition, and when the confidence coefficient is smaller than a preset threshold value, taking a corresponding target license plate character image as the input of a character recognition model, and acquiring a recognition result output by the character recognition model, wherein the character recognition model is obtained by adopting deep learning algorithm training.
FIG. 9 is a diagram illustrating an internal structure of a computer device in one embodiment. The computer device may specifically be a server and a terminal device, where the server includes, but is not limited to, a high-performance computer and a high-performance computer cluster; the terminal devices include, but are not limited to, mobile terminal devices including, but not limited to, mobile phones, tablet computers, smart watches, and laptops, and desktop terminal devices including, but not limited to, desktop computers and in-vehicle computers. As shown in fig. 9, the computer device includes a processor, a memory, and a network interface connected by a system bus. The memory comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program that, when executed by the processor, causes the processor to implement the license plate recognition method. The internal memory may also have a computer program stored therein that, when executed by the processor, causes the processor to perform a license plate recognition method. Those skilled in the art will appreciate that the architecture shown in fig. 9 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, the license plate recognition method provided by the present application may be implemented in the form of a computer program, and the computer program may be run on a computer device as shown in fig. 9. The memory of the computer device can store various program templates which form the license plate recognition device. For example, the obtaining module 602, the detecting module 604, the correcting module 606, and the identifying module 608.
A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing the following steps when executing the computer program: detecting an image containing a license plate to obtain a detected candidate license plate region; taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model; correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image; and identifying license plate characters in the target license plate image to obtain an identification result.
In one embodiment, the license plate vertex detection model includes: a first detection model, a second detection model and a third detection model; the step of taking the candidate license plate region as the input of a license plate vertex detection model and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model comprises the following steps: taking the candidate license plate region as the input of the first detection model, wherein the first detection model is used for detecting the candidate license plate region to generate a candidate license plate window; acquiring a candidate license plate window output by the first detection model, and correcting the candidate license plate window to obtain a first candidate license plate window; taking the first candidate license plate window as the input of the second detection model, wherein the second detection model is used for screening the first candidate license plate window; correcting the candidate license plate window output by the second detection model to obtain a second candidate license plate window; and taking the second candidate license plate window as the input of the third detection model, wherein the third detection model is used for identifying license plate areas to obtain the positions of four license plate vertexes.
In one embodiment, the obtaining of the candidate license plate window output by the first detection model and the correction of the candidate license plate window to obtain a first candidate license plate window include: obtaining a candidate license plate window and a license plate frame regression vector output by the first detection model; and correcting the license plate window by adopting the license plate frame regression vector to obtain the first candidate license plate window.
In one embodiment, the correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image includes: calculating according to the positions of the four license plate vertexes to obtain a perspective transformation matrix; and correcting the license plate image according to the perspective transformation matrix to obtain a target license plate image.
In one embodiment, the detecting an image including a license plate to obtain a detected candidate license plate region includes: acquiring an original image containing a license plate, and preprocessing the original image, wherein the preprocessing comprises the following steps: at least one of gray scale processing, scaling processing and denoising processing; carrying out vertical edge detection on the preprocessed image, and determining the height and width of the image; dividing the image into a plurality of image blocks according to the height and the width of the image; determining a comparison value corresponding to each image block according to pixel values in the image blocks; carrying out binarization processing on the image blocks according to the comparison value of each image block and a preset binarization threshold value to obtain a binarization image; and detecting the license plate region in the binary image to obtain a candidate license plate region.
In one embodiment, the detecting the license plate region in the binarized image to obtain a candidate license plate region includes: carrying out corrosion expansion processing on each detected region by adopting a morphological algorithm on the binary image to obtain an image containing a communicated region; conducting connected region search on the images containing the connected regions to obtain suspected license plate regions; and detecting according to the texture characteristics of each suspected license plate area, and removing the pseudo license plate areas to obtain candidate license plate areas.
In one embodiment, the recognizing characters in the target license plate image to obtain a recognition result includes: detecting and segmenting characters in the target license plate image to obtain license plate character images; carrying out normalization processing on the license plate character image to obtain a target license plate character image; extracting the characteristics of the target license plate character image, and recognizing characters according to the extracted characteristics; and acquiring a confidence coefficient of character recognition, and when the confidence coefficient is smaller than a preset threshold value, taking a corresponding target license plate character image as the input of a character recognition model, and acquiring a recognition result output by the character recognition model, wherein the character recognition model is obtained by adopting deep learning algorithm training.
A computer-readable storage medium, in which a computer program is stored which, when executed by a processor, carries out the steps of: detecting an image containing a license plate to obtain a detected candidate license plate region; taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertexes obtained by detection output by the license plate vertex detection model; correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image; and identifying license plate characters in the target license plate image to obtain an identification result.
In one embodiment, the license plate vertex detection model comprises: a first detection model, a second detection model and a third detection model; the step of taking the candidate license plate region as the input of a license plate vertex detection model and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model comprises the following steps: taking the candidate license plate region as the input of the first detection model, wherein the first detection model is used for detecting the candidate license plate region to generate a candidate license plate window; acquiring a candidate license plate window output by the first detection model, and correcting the candidate license plate window to obtain a first candidate license plate window; taking the first candidate license plate window as the input of the second detection model, wherein the second detection model is used for screening the first candidate license plate window; correcting the candidate license plate window output by the second detection model to obtain a second candidate license plate window; and taking the second candidate license plate window as the input of the third detection model, wherein the third detection model is used for identifying license plate areas to obtain the positions of four license plate vertexes.
In one embodiment, the obtaining of the candidate license plate window output by the first detection model, and performing correction processing on the candidate license plate window to obtain a first candidate license plate window includes: obtaining a candidate license plate window and a license plate frame regression vector output by the first detection model; and correcting the license plate window by adopting the license plate frame regression vector to obtain the first candidate license plate window.
In one embodiment, the correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image includes: calculating according to the positions of the four license plate vertexes to obtain a perspective transformation matrix; and correcting the license plate image according to the perspective transformation matrix to obtain a target license plate image.
In one embodiment, the detecting an image including a license plate to obtain a detected candidate license plate region includes: acquiring an original image containing a license plate, and preprocessing the original image, wherein the preprocessing comprises the following steps: at least one of gray scale processing, scaling processing and denoising processing; carrying out vertical edge detection on the preprocessed image, and determining the height and width of the image; dividing the image into a plurality of image blocks according to the height and the width of the image; determining a comparison value corresponding to each image block according to pixel values in the image blocks; carrying out binarization processing on the image blocks according to the comparison value of each image block and a preset binarization threshold value to obtain a binarization image; and detecting the license plate region in the binary image to obtain a candidate license plate region.
In one embodiment, the detecting the license plate region in the binarized image to obtain a candidate license plate region includes: carrying out corrosion expansion processing on each detected region on the binary image by adopting a morphological algorithm to obtain an image containing a communicated region; conducting connected region search on the images containing the connected regions to obtain suspected license plate regions; and detecting according to the texture characteristics of each suspected license plate area, and removing the pseudo license plate areas to obtain candidate license plate areas.
In one embodiment, the recognizing characters in the target license plate image to obtain a recognition result includes: detecting and segmenting characters in the target license plate image to obtain license plate character images; carrying out normalization processing on the license plate character image to obtain a target license plate character image; extracting the characteristics of the target license plate character image, and recognizing the characters according to the extracted characteristics; and acquiring a confidence coefficient of character recognition, and when the confidence coefficient is smaller than a preset threshold value, taking a corresponding target license plate character image as the input of a character recognition model, and acquiring a recognition result output by the character recognition model, wherein the character recognition model is obtained by adopting deep learning algorithm training.
It should be noted that the license plate recognition method, the license plate recognition apparatus, the computer device, and the computer-readable storage medium described above belong to a general inventive concept, and the contents in the embodiments of the license plate recognition method, the license plate recognition apparatus, the computer device, and the computer-readable storage medium may be mutually applicable.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the program is executed. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), synchronous Link (Synchlink) DRAM (SLDRAM), Rambus (Rambus) direct RAM (RDRAM), direct bused dynamic RAM (DRDRAM), and bused dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, and these are all within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims. Please enter the implementation details.

Claims (6)

1. A license plate recognition method is characterized by comprising the following steps:
detecting an image containing a license plate to obtain a detected candidate license plate region;
taking the candidate license plate region as the input of a license plate vertex detection model, and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model;
correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image; recognizing the license plate characters in the target license plate image to obtain a recognition result,
the detecting the image containing the license plate to obtain the detected candidate license plate region comprises the following steps: acquiring an original image containing a license plate, and preprocessing the original image, wherein the preprocessing comprises the following steps: at least one of gray processing, scaling processing and denoising processing; carrying out vertical edge detection on the preprocessed image, and determining the height and width of the image; dividing the image into a plurality of image blocks according to the height and the width of the image; determining a comparison value corresponding to each image block according to the pixel values in the image blocks; carrying out binarization processing on the image blocks according to the comparison value of each image block and a preset binarization threshold value to obtain a binarization image; the binarization threshold is to sum the pixel values of each pixel in the image block, then calculate an average value, use the average value as the binarization threshold of the image block, set the pixels larger than the binarization threshold to be 255, and otherwise set to be 0; or the binarization threshold is to obtain a maximum pixel value and a minimum pixel value in each image block, and average the maximum pixel value and the minimum pixel value to obtain a corresponding comparison value, and the calculation of the comparison value can be represented by the following formula:
Thr i ＝(max+min)/2
wherein max is the maximum pixel value in the region, min is the minimum pixel value in the region, and then each point in the region is binarized by using the following formula:
Figure 186733DEST_PATH_IMAGE001
b is the value of each pixel point of the binary image, v is the pixel value of the midpoint of the image, Thr i The comparison value of the block to which the pixel point belongs, T is a threshold value of binaryzation;
detecting the license plate region in the binary image to obtain a candidate license plate region; the detecting the license plate region in the binary image to obtain a candidate license plate region comprises the following steps: carrying out corrosion expansion processing on each detected region by adopting a morphological algorithm on the binary image to obtain an image containing a communicated region; conducting connected region search on the images containing the connected regions to obtain suspected license plate regions; detecting according to the texture characteristics of each suspected license plate area, and removing the pseudo license plate areas to obtain candidate license plate areas; the texture feature identification adopts the following mode: carrying out binarization processing on the suspected license plate area, then counting the number of lines L with jump larger than a set threshold value in the area, and removing the suspected license plate area if the number of lines L is smaller than a preset threshold value T, wherein the preset threshold value T can be half of the height of the area;
the recognizing the characters in the target license plate image to obtain a recognition result comprises the following steps: detecting and segmenting characters in the target license plate image to obtain license plate character images; carrying out normalization processing on the license plate character image to obtain a target license plate character image; extracting the characteristics of the target license plate character image, and recognizing characters according to the extracted characteristics; acquiring a confidence coefficient of character recognition, and when the confidence coefficient is smaller than a preset threshold value, inputting a corresponding target license plate character image as a character recognition model, and acquiring a recognition result output by the character recognition model, wherein the character recognition model is obtained by adopting deep learning algorithm training, the segmented characters can be unified to the size of 24 pixels with the width of 48 and the height, Gabor characteristics of the character image are extracted, an SVM is adopted for character recognition, when the confidence coefficient of the character recognition is smaller than the set threshold value, the character recognition model is adopted for secondary recognition of the character, and then the result is output;
the license plate vertex detection model comprises: a first detection model, a second detection model and a third detection model; the first detection model is a model composed of convolutional layers, and includes: the input image has a width of 16 and a height of 12, and the convolution kernel of each convolution layer adopts 3X 3; the second detection model is also composed of convolution layers, the width of an image input by the second detection model is 36, the height of the image input by the second detection model is 24, and the second detection model comprises the following components: the convolution kernels of the first convolution layer and the second convolution layer are 3X3, and the convolution kernels of the third convolution layer are 2X 2; the third detection model includes: four convolutional layers and 1 full-link layer, the convolutional kernels of the first convolutional layer, the second convolutional layer and the third convolutional layer are 3X3, the convolutional kernel of the fourth convolutional layer is 2X2, the width of the input image of the third detection model is 48, and the height is 72; the step of taking the candidate license plate region as the input of a license plate vertex detection model and acquiring the positions of four license plate vertices obtained by detection output by the license plate vertex detection model comprises the following steps: taking the candidate license plate region as the input of the first detection model, wherein the first detection model is used for detecting the candidate license plate region to generate a candidate license plate window; acquiring a candidate license plate window output by the first detection model, performing correction processing on the candidate license plate window, and merging the candidate license plate windows with high overlapping rate by adopting NMS (network management system) on the candidate license plate windows after the correction processing to obtain a first candidate license plate window; taking the first candidate license plate window as the input of the second detection model, wherein the second detection model is used for screening the first candidate license plate window; correcting the candidate license plate windows output by the second detection model, and combining the candidate license plate windows with high overlapping rate by adopting NMS (network management system) to obtain a second candidate license plate window; and taking the second candidate license plate window as the input of the third detection model, wherein the third detection model is used for identifying license plate areas to obtain the positions of four license plate vertexes.
2. The method of claim 1, wherein the obtaining of the candidate license plate window output by the first detection model and the correction of the candidate license plate window to obtain a first candidate license plate window comprises: obtaining a candidate license plate window and a license plate frame regression vector output by the first detection model; and correcting the license plate window by adopting the license plate frame regression vector to obtain the first candidate license plate window.
3. The method of claim 1, wherein the correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image comprises: calculating according to the positions of the four license plate vertexes to obtain a perspective transformation matrix; and correcting the license plate image according to the perspective transformation matrix to obtain a target license plate image.
4. A license plate recognition apparatus for realizing the license plate recognition method according to any one of claims 1 to 3, the apparatus comprising: the acquisition module is used for detecting an image containing a license plate and acquiring a detected candidate license plate area; the detection module is used for taking the candidate license plate area as the input of a license plate vertex detection model and acquiring the positions of four license plate vertexes obtained by detection output by the license plate vertex detection model; the correction module is used for correcting the license plate image according to the positions of the four license plate vertexes to obtain a corrected target license plate image; and the recognition module is used for recognizing the license plate characters in the target license plate image to obtain a recognition result.
5. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the license plate recognition method according to any one of claims 1 to 3 when executing the computer program.
6. A computer-readable storage medium, in which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of a license plate recognition method according to any one of claims 1 to 3.
CN201910626156.3A 2019-07-11 2019-07-11 License plate recognition method and device, computer equipment and storage medium Active CN110414507B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910626156.3A CN110414507B (en) 2019-07-11 2019-07-11 License plate recognition method and device, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910626156.3A CN110414507B (en) 2019-07-11 2019-07-11 License plate recognition method and device, computer equipment and storage medium

Publications (2)

Publication Number Publication Date
CN110414507A CN110414507A (en) 2019-11-05
CN110414507B true CN110414507B (en) 2022-07-26

Family

ID=68361096

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910626156.3A Active CN110414507B (en) 2019-07-11 2019-07-11 License plate recognition method and device, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN110414507B (en)

Families Citing this family (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111079744B (en) * 2019-12-06 2020-09-01 鲁东大学 Intelligent vehicle license plate identification method and device suitable for complex illumination environment
CN111126383A (en) * 2019-12-06 2020-05-08 广州烽火众智数字技术有限公司 License plate detection method, system, device and storage medium
CN111191653A (en) * 2019-12-25 2020-05-22 北京精英路通科技有限公司 License plate recognition method and device, computer equipment and storage medium
CN111340045B (en) * 2020-02-12 2023-09-01 浙江大华技术股份有限公司 License plate number recognition method, device and storage medium
CN113326836A (en) * 2020-02-28 2021-08-31 深圳市丰驰顺行信息技术有限公司 License plate recognition method and device, server and storage medium
CN111695561A (en) * 2020-05-25 2020-09-22 南京博雅集智智能技术有限公司 License plate detection and correction recognition method and recognition system based on SSD
CN111914834B (en) * 2020-06-18 2024-04-02 绍兴埃瓦科技有限公司 Image recognition method, device, computer equipment and storage medium
CN112434700A (en) * 2020-11-25 2021-03-02 创新奇智(上海)科技有限公司 License plate recognition method, device, equipment and storage medium
CN112580648A (en) * 2020-12-14 2021-03-30 成都中科大旗软件股份有限公司 Method for realizing image information identification based on image segmentation technology
CN112634141B (en) * 2020-12-23 2024-03-29 浙江大华技术股份有限公司 License plate correction method, device, equipment and medium
CN112560754A (en) * 2020-12-23 2021-03-26 北京百度网讯科技有限公司 Bill information acquisition method, device, equipment and storage medium
CN112686252A (en) * 2020-12-28 2021-04-20 中国联合网络通信集团有限公司 License plate detection method and device
CN113095320A (en) * 2021-04-01 2021-07-09 湖南大学 License plate recognition method and system and computing device
CN112926583B (en) * 2021-04-25 2022-08-16 南京甄视智能科技有限公司 License plate recognition method and license plate recognition system
CN113313124B (en) * 2021-07-29 2021-11-05 佛山市墨纳森智能科技有限公司 Method and device for identifying license plate number based on image segmentation algorithm and terminal equipment
CN115019297B (en) * 2022-08-04 2022-12-09 之江实验室 Real-time license plate detection and identification method and device based on color augmentation
CN117409376B (en) * 2023-12-15 2024-05-10 南京中鑫智电科技有限公司 Infrared online monitoring method and system for high-voltage sleeve

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104392205A (en) * 2014-10-24 2015-03-04 浙江力石科技股份有限公司 Abnormal vehicle license plate recognition method and system
CN107679531A (en) * 2017-06-23 2018-02-09 平安科技（深圳）有限公司 Licence plate recognition method, device, equipment and storage medium based on deep learning

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101408977B (en) * 2008-11-24 2012-04-18 东软集团股份有限公司 Method and apparatus for dividing candidate barrier region
CN102156868B (en) * 2011-03-31 2013-03-13 汉王科技股份有限公司 Image binaryzation method and device
CN102163284B (en) * 2011-04-11 2013-02-27 西安电子科技大学 Chinese environment-oriented complex scene text positioning method
CN102426649B (en) * 2011-10-13 2013-08-21 石家庄开发区冀科双实科技有限公司 Simple steel seal digital automatic identification method with high accuracy rate
CN103093181B (en) * 2011-11-01 2016-04-27 青岛海信网络科技股份有限公司 A kind of method and apparatus of license plate image location
CN103065138B (en) * 2012-12-06 2015-07-15 中通服公众信息产业股份有限公司 Recognition method of license plate number of motor vehicle
CN104050450A (en) * 2014-06-16 2014-09-17 西安通瑞新材料开发有限公司 Vehicle license plate recognition method based on video
CN105260701B (en) * 2015-09-14 2019-01-29 中电海康集团有限公司 A kind of front vehicles detection method suitable under complex scene
CN105320953A (en) * 2015-09-28 2016-02-10 万永秀 License plate recognition method
US9785855B2 (en) * 2015-12-17 2017-10-10 Conduent Business Services, Llc Coarse-to-fine cascade adaptations for license plate recognition with convolutional neural networks
CN107229929A (en) * 2017-04-12 2017-10-03 西安电子科技大学 A kind of license plate locating method based on R CNN
CN107506763B (en) * 2017-09-05 2020-12-01 武汉大学 Multi-scale license plate accurate positioning method based on convolutional neural network

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104392205A (en) * 2014-10-24 2015-03-04 浙江力石科技股份有限公司 Abnormal vehicle license plate recognition method and system
CN107679531A (en) * 2017-06-23 2018-02-09 平安科技（深圳）有限公司 Licence plate recognition method, device, equipment and storage medium based on deep learning

Also Published As

Publication number Publication date
CN110414507A (en) 2019-11-05

Similar Documents

Publication Publication Date Title
CN110414507B (en) License plate recognition method and device, computer equipment and storage medium
CN108898610B (en) Object contour extraction method based on mask-RCNN
CN108960211B (en) Multi-target human body posture detection method and system
CN112686812B (en) Bank card inclination correction detection method and device, readable storage medium and terminal
JP5775225B2 (en) Text detection using multi-layer connected components with histograms
CN108268867B (en) License plate positioning method and device
CN109711416B (en) Target identification method and device, computer equipment and storage medium
CN108986152B (en) Foreign matter detection method and device based on difference image
CN108197644A (en) A kind of image-recognizing method and device
CN114529459B (en) Method, system and medium for enhancing image edge
CN111680690B (en) Character recognition method and device
CN109492642B (en) License plate recognition method, license plate recognition device, computer equipment and storage medium
CN111461170A (en) Vehicle image detection method and device, computer equipment and storage medium
CN110598788A (en) Target detection method and device, electronic equipment and storage medium
CN109447117B (en) Double-layer license plate recognition method and device, computer equipment and storage medium
CN108960247B (en) Image significance detection method and device and electronic equipment
WO2022121021A1 (en) Identity card number detection method and apparatus, and readable storage medium and terminal
CN111709377B (en) Feature extraction method, target re-identification method and device and electronic equipment
CN110751623A (en) Joint feature-based defect detection method, device, equipment and storage medium
CN114627456A (en) Bill text information detection method, device and system
CN111640071A (en) Method for obtaining panoramic foreground target based on convolutional neural network frame difference repairing method
CN112926610A (en) Construction method of license plate image screening model and license plate image screening method
CN115984316B (en) Industrial image edge extraction method and device for complex environment
CN113743413B (en) Visual SLAM method and system combining image semantic information
CN112862802B (en) Location recognition method based on edge appearance sequence matching

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
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20210430

Address after: A1926, building 9, zone 2, Shenzhen Bay science and technology ecological park, 3609 Baishi Road, high tech Zone community, Yuehai street, Nanshan District, Shenzhen, Guangdong 518000

Applicant after: Shenzhen zhiyouting Technology Co.,Ltd.

Address before: 518000 2nd floor, No.5 Zhongxing Road, guantian community, Shiyan street, Bao'an District, Shenzhen City, Guangdong Province

Applicant before: HECHANG FUTURE TECHNOLOGY (SHENZHEN) Co.,Ltd.

GR01 Patent grant
GR01 Patent grant