WO2019174405A1 - 车牌辨识方法以及其系统 - Google Patents

车牌辨识方法以及其系统 Download PDF

Info

Publication number
WO2019174405A1
WO2019174405A1 PCT/CN2019/072542 CN2019072542W WO2019174405A1 WO 2019174405 A1 WO2019174405 A1 WO 2019174405A1 CN 2019072542 W CN2019072542 W CN 2019072542W WO 2019174405 A1 WO2019174405 A1 WO 2019174405A1
Authority
WO
WIPO (PCT)
Prior art keywords
license plate
image
plate recognition
character
feature
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.)
Ceased
Application number
PCT/CN2019/072542
Other languages
English (en)
French (fr)
Inventor
陈昱达
梁峰铭
郑景鸿
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.)
Delta Electronics Inc
Original Assignee
Delta Electronics Inc
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 Delta Electronics Inc filed Critical Delta Electronics Inc
Priority to US16/980,747 priority Critical patent/US11443535B2/en
Priority to JP2020548723A priority patent/JP7044898B2/ja
Priority to EP19768532.4A priority patent/EP3767530A4/en
Publication of WO2019174405A1 publication Critical patent/WO2019174405A1/zh
Anticipated expiration legal-status Critical
Priority to US17/881,218 priority patent/US20220375236A1/en
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • G06V10/449—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
    • G06V10/451—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters with interaction between the filter responses, e.g. cortical complex cells
    • G06V10/454—Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition
    • G06V30/14—Image acquisition
    • G06V30/148—Segmentation of character regions
    • G06V30/153—Segmentation of character regions using recognition of characters or words
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition
    • G06V30/19—Recognition using electronic means
    • G06V30/191—Design or setup of recognition systems or techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
    • G06V30/1916—Validation; Performance evaluation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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

Definitions

  • the invention relates to a license plate recognition method and a license plate recognition system, in particular to a license plate recognition method and a license plate recognition system for recognizing each character of a license plate by using a neural network.
  • license plate recognition technology In the application of image processing, license plate recognition technology has been widely known.
  • common technical means for obtaining license plate information are license plate location, license plate character cutting and license plate character recognition.
  • the license plate image may have obvious license plate characteristics, license plate skew, license plate deformation, light noise and license plate breakage. This leads to a decrease in the accuracy of identification.
  • the existing license plate location technology usually finds the location of the license plate image based on the edge density value. If the license plate is stained or decorated, the edge density value may be destroyed, resulting in a significant decrease in the accuracy of the license plate location.
  • An embodiment of the present invention provides a license plate recognition method, including the steps of: acquiring a to-be-processed image including all license plate characters; and extracting, by a feature map extraction module, a plurality of feature maps having the character features of the image to be processed; A block corresponding to each character block and coordinates are retrieved according to the feature map based on the character network identification model based on the neural network; and a license plate recognition result is obtained according to the block of each character and the coordinates.
  • Another embodiment of the present invention further provides a license plate recognition system including an image capture unit and a processing unit.
  • the image capturing unit is configured to capture at least one original image.
  • the processing unit is configured to: receive an original image from the image capturing unit; obtain a to-be-processed image including all the license plate characters according to the original image; and extract a plurality of feature maps having the character features of the image to be processed by using a feature map extraction model; The block and the coordinates corresponding to each character are retrieved according to the feature map by the one-character identification model based on the neural network; and a license plate recognition result is obtained according to the block and the coordinates of each character.
  • Another embodiment of the present invention further provides a license plate recognition method, comprising the steps of: obtaining a to-be-processed image; acquiring a plurality of feature maps having a plurality of target features by a feature map extraction module; obtaining each through a target location extraction module At least one region of the feature map having the target feature, and each frame of each feature map is assigned a score corresponding to each target feature; each frame of each of the above feature maps is performed according to the score by a target candidate classification module Classifying and retaining at least one region corresponding to the character feature; and obtaining a license plate recognition result from the region corresponding to the character feature by a voting/statistics module.
  • FIG. 1 is a system architecture diagram showing a license plate recognition system according to an embodiment of the invention.
  • FIG. 2 is a flow chart showing a method of identifying a license plate according to an embodiment of the invention.
  • FIG. 3A is a schematic diagram showing a current image according to an embodiment of the invention.
  • FIG. 3B is a schematic diagram showing image changes of a current image and a historical background image according to an embodiment of the invention.
  • 4A-4D are schematic diagrams showing trained matrices for generating feature maps in accordance with some embodiments of the present invention.
  • 5A and 5B are schematic diagrams showing a block judged to have a character according to an embodiment of the present invention.
  • FIG. 6 is a schematic diagram showing a voting/statistics module according to an embodiment of the invention.
  • FIG. 7A is a schematic diagram showing a current image according to an embodiment of the invention.
  • FIG. 7B is a schematic diagram showing a vehicle head image according to an embodiment of the invention.
  • FIG. 8 is a schematic diagram showing a license plate text area according to an embodiment of the invention.
  • FIG. 9 is a flow chart showing a license plate recognition method according to another embodiment of the present invention.
  • the license plate recognition system 100 can be implemented in an electronic device such as a desktop computer, a notebook computer, or a tablet computer, and the license plate recognition system 100 includes at least one processing unit 110.
  • the processing unit 110 can be implemented in various manners, such as in dedicated hardware circuits or general purpose hardware (eg, a single processor, a multiprocessor with parallel processing capabilities, a graphics processor, or other processor capable of computing), and The functions described later are provided when the program code or software related to each model and process of the present invention is provided.
  • the license plate recognition system 100 further includes a storage unit 120 for storing the acquired images, data required for execution, and various electronic files, such as various algorithms and/or models.
  • the license plate recognition system 100 further includes an image capture unit 130, such as a monitor, a camera, and/or a camera, for acquiring at least one image or a continuous video image and transmitting it back to the processing unit 110.
  • the display unit 140 can be a display panel (for example, a thin film liquid crystal display panel, an organic light emitting diode panel, or other display capable panel) for displaying input characters, numbers, symbols, moving tracks of the drag mouse, or provided by an application. User interface for viewing to the user.
  • the license plate recognition system 100 further includes an input device (not shown) such as a mouse, a stylus or a keyboard for the user to perform a corresponding operation.
  • step S201 the image capturing unit 130 obtains a to-be-processed image.
  • the processing unit 110 can compare the current image with the historical image to determine whether the current image is in the current image. There are vehicles or other objects entering the shooting range.
  • the processing unit 110 can obtain a historical background image through a background subtract module according to the plurality of historical images, so that the processing unit 110 can quickly determine the historical background image and the current image.
  • FIG. 3A is a schematic diagram of a current image
  • FIG. 3B is a schematic diagram of image changes between a current image and a historical background image.
  • the area of the image change is about 37% according to the content of FIG. 3B. If the predetermined value is set to 35%, the processing unit 110 can determine that a vehicle or other object appears in the current image.
  • step S202 the processing unit 110 receives the image to be processed, and obtains a plurality of feature maps through a feature map extraction module.
  • the feature map extraction module can be trained by strengthening the matrix of character features, which is mainly used to highlight characters such as English letters or numbers in the image.
  • 4A-4D are schematic diagrams showing a plurality of trained matrices for acquiring a feature map, in accordance with some embodiments of the present invention.
  • step S203 after acquiring the feature map, the processing unit 110 retrieves the block corresponding to each character and the corresponding coordinates according to the feature map by using a character recognition model.
  • the character recognition module is a neural network-based module, which mainly uses a plurality of images corresponding to various letters (ie, A to Z) and numbers (ie, 0 to 9) as training data to accurately recognize the image.
  • the plurality of blocks shown in FIG. 5A are blocks that are determined to have characters.
  • the character recognition module can directly find the block corresponding to each character, without first dividing the characters in the license plate.
  • the processing unit 110 further selects the reliability.
  • Higher regions eg, by selecting regions with more overlap
  • regions with more overlap serve as a basis for ordering the license plate characters.
  • the area selected by the thick frame line is the last extracted area having characters, and the previously determined block can be effectively filtered by the above method.
  • the processing unit 110 obtains a license plate recognition result according to the order of each character and coordinates after acquiring all the characters and their corresponding coordinates.
  • the processing unit 110 can vote on a plurality of license plate images through a voting/statistics module to improve the accuracy of the license plate recognition result.
  • the processing unit 110 may classify the license plate area into at least two groups by the license plate grouping rule.
  • the license plate grouping rules may include a license plate naming grouping rule, an English word area and a digital word grouping rule, a dash grouping rule, and a character relative position grouping rule.
  • FIG. 6 is a schematic diagram showing a voting/statistics module according to an embodiment of the invention.
  • the processing unit 110 divides the license plate into two groups (ie, the left half and the right half of the dash) based on the position of the dash using the dash grouping rule. Then, after obtaining the identification results of the two subgroups, the different identification results in each subgroup are voted, and if there are repeated identification results, the voting scores are accumulated. For example, as shown in FIG.
  • the processing unit 110 further assigns different weights to the recognition result according to the time order of the identification result. For example, newer recognition results give greater weight, while older identification results give less weight, thereby speeding up the convergence of the final license plate recognition results.
  • the processing unit 110 may further adopt a head image capturing module or a rear image.
  • the module takes a front image or a rear image in the current image to reduce the area of the image to be processed.
  • the front image capturing module or the rear image capturing module trains various head images or vehicles through multiple image features (such as Haar Feature, HOG, LBP, etc.) with a classifier (Cascade Classifier, Ada boost or SVM).
  • the tail image to obtain the front image or the rear image from the current image.
  • FIG. 7A is a schematic diagram showing a current image
  • FIG. 7B is a schematic diagram showing a front image obtained by the front image capturing module.
  • the processing unit 110 may detect the model image from the front of the vehicle through a license plate character area or The area near the license plate is obtained from the rear image.
  • the license plate character area detection model also trains various character images by using multiple image features (such as Haar Feature, HOG, LBP, etc.) with a classifier (Cascade Classifier, Ada boost or SVM) to capture images from the front or the rear of the car. Find each character in the image. For example, as shown in FIG.
  • the processing unit 110 detects the model by the license plate character region to find four regions 801-804 having characters from the front image. Next, the processing unit 110 combines the regions 801-804 to obtain another larger region 810, and expands according to the format enlargement region 810 of the license plate to obtain another region including all the license plate characters.
  • the license plate has six characters, and the license plate character area detection model only finds four areas with characters from the front image, so in order to ensure that all characters in the license plate are included in the image to be processed, processing The unit 110 can further determine the outward expansion magnification according to the number of characters found. For example, as shown in FIG.
  • processing unit 110 since processing unit 110 has found four characters, processing unit 110 expands approximately twice from four sides of region 810 (as indicated by region 820), thus ensuring All characters are included in the image to be processed. In other words, if the processing unit 110 only finds an area with a character, the processing unit 110 adaptively increases the magnification of the expansion (eg, ten times to the left and right) to ensure that all characters are included. Process the image.
  • the expansion ratio can be adjusted according to the needs of the user.
  • the foregoing embodiments are only used for illustrative purposes, and the invention is not limited thereto. Compared with the front image or the rear image, the image to be processed obtained by the license plate character area detection model more precisely reduces the area of the image to further improve the operation speed.
  • the license plate character area detection model since the main function of the license plate character area detection model is only to find the area that may have characters, not to accurately identify the characters, the license plate character area detection is compared with the character recognition model.
  • the model belongs to the weak classifier, that is, its detection accuracy is low but the calculation speed is faster.
  • the license plate character area detection model and the front image capturing module or the rear image capturing module use different image features and classifiers.
  • step S205 in order to further improve the accuracy of the character recognition model, the processing unit 110 further uses each image and the corresponding recognition result as training data to update the character recognition model.
  • the above identification result includes the correct license plate recognition result and the incorrect license plate recognition result, thereby reducing the identification error of the character recognition module and indirectly speeding up the processing speed of the license plate recognition system.
  • FIG. 9 is a flow chart showing a license plate recognition method according to another embodiment of the present invention.
  • the image capturing unit 130 obtains at least one image to be processed.
  • the processing unit 110 receives the acquired image to be processed from the image capturing unit 130, and acquires a plurality of feature maps through the feature map extraction module.
  • the information contained in the feature map includes a plurality of target features corresponding to different spatial frequencies (eg, from low frequency to high frequency), and the target features may include character features on the license plate, license plate appearance features, background features, vehicles Information characteristics (such as directional mirrors, models, wheels, etc., representing the characteristics of the car).
  • the feature map extraction module can be trained by a matrix containing the aforementioned target features.
  • the processing unit 110 finds the region having the target feature by using a target location extraction module according to the feature map.
  • the processing unit 110 may extract a frame for each predetermined pixel on the feature map by means of a clustering manner or a customized size, and determine, according to the feature map extraction module, features that may be included in each frame. And give each box a score corresponding to each target feature type.
  • the processing unit 110 first obtains a target sensitive score map of the feature map by using a simple classifier, that is, finding a plurality of target feature points or target feature regions having the target feature on the feature map, and then A plurality of regions located near the target feature point are selected using frame rings having different sizes, and the regions are given scores corresponding to each of the target features.
  • step S904 the processing unit 110 retains only the highest score by a target candidate classification module in a non-maximum value suppression manner.
  • a target feature with a score greater than a predetermined value to classify the position corresponding to each frame. For example, if a certain box corresponds to the largest score of the background feature and is greater than the predetermined value, the processing unit 110 classifies the frame as a frame corresponding to the background feature. In addition, when a certain box corresponds to a score of each target feature that is not greater than a predetermined value, the area is classified as a non-target feature.
  • the processing unit 110 can further group the plurality of frames having the same target feature and adjacent to each other into a larger area by the target candidate classification module, so as to facilitate the subsequent identification process.
  • the processing unit 110 retains only the area corresponding to the character feature, and proceeds to step S905.
  • the processing unit 110 obtains a license plate recognition result according to each character and the order of the coordinates (for example, from left to right, from top to bottom).
  • the processing unit 110 can vote on a plurality of license plate images through the aforementioned voting/statistics module to improve the accuracy of the license plate recognition result.
  • the license plate recognition method described in step S905 is similar to the license plate recognition method in step S204, and is not described here to simplify the description.
  • step S906 the processing unit 110 further uses each image to be processed and the corresponding recognition result as training data to update the character recognition model.
  • the above identification result includes the correct license plate recognition result and the incorrect license plate recognition result, thereby reducing the identification error of the character recognition module.
  • the license plate recognition method and the license plate recognition system can still be performed in an environment with poor viewing angle or complicated change through the aforementioned license plate image capturing step and license plate character recognition step. Maintaining rapid identification speed and high accuracy, and continuously using the identification results as training data, can further reduce the error of license plate recognition and indirectly accelerate the calculation speed of the license plate recognition system.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Character Discrimination (AREA)
  • Traffic Control Systems (AREA)
  • Image Analysis (AREA)
  • Character Input (AREA)

Abstract

一种车牌辨识方法，包括以下步骤：取得包含所有车牌字元的一待处理影像；通过一特征地图提取模块强化提取对应于上述待处理影像的字元特征的多个特征地图；通过基于神经网络的一字元辨识模型根据上述特征地图撷取对应于每个字元的区块以及坐标；以及根据每个字元的上述区块以及上述坐标取得一车牌辨识结果。

Description

车牌辨识方法以及其系统 技术领域
本发明涉及一种车牌辨识方法以及车牌辨识系统，特别涉及一种利用神经网络辨识车牌的每个字元的车牌辨识方法以及车牌辨识系统。
背景技术
在影像处理的应用中，车牌辨识技术已被广为人知。而于现有的车牌辨识技术中，取得车牌信息的常用技术手段为车牌定位、车牌字元切割以及车牌字元辨识。然而，于实际应用上，因为拍摄角度不同或是光源、日夜、晴雨等环境的干扰，将使得车牌影像将可能出现车牌特征不明显、车牌歪斜、车牌变形、光噪及车牌断裂的情况，进而造成辨识准确率下降。此外，现有的车牌定位技术通常是基于边缘密度值寻找车牌影像位置，若车牌有污损、装饰等情况将可能破坏边缘密度值的特征，导致车牌定位的正确率大幅度下降。再者，若取得的车牌过于歪斜或者变形亦会导致字元切割难以执行，必须通过额外的演算法对车牌进行校正。以上各种因素显示现有的车牌辨识技术对于环境的容忍度低，必须要通过各种额外的影像处理技术以提高辨识率，然此举亦会降低车牌辨识的速度。因此，要如何提供更佳的车牌辨识方法以提高车牌辨识对环境的容忍度并维持高准确率以及快的辨识速度，为目前必须解决的问题。
发明内容
本发明一实施例提供一种车牌辨识方法，包括以下步骤：取得包含所有车牌字元的一待处理影像；通过一特征地图提取模块提取具有上述待处理影像的字元特征的多个特征地图；通过基于神经网络的一字元辨识模型根据上述特征地图撷取对应于每个字元的区块以及坐标；以及根据每个字元的上述区块以及上述坐标取得一车牌辨识结果。
本发明另一实施例更提供一种车牌辨识系统，包括一影像撷取单元以及一处理单元。影像撷取单元用以撷取至少一原始影像。处理单元用以：自影像撷取单元接收原始影像；根据原始影像取得包含所有车牌字元的一待处理影像；通过一特征地图提取模型提取具有待处理影像的字元特征的多个特征地图；通过基于神经网络的一字元辨识模型根据特征地 图撷取对应于每个字元的区块以及坐标；以及根据每个字元的区块以及坐标取得一车牌辨识结果。
本发明另一实施例更提供一种车牌辨识方法，包括以下步骤：取得一待处理影像；通过一特征地图提取模块取得具有多个目标特征的多个特征地图；通过一目标位置提取模块取得每个特征地图中具有目标特征的至少一区域，并给予每个特征地图的每个框对应于每个目标特征的分数；通过一目标候选分类模块根据分数对每个上述特征地图的每个框进行分类，并保留对应于文字特征的至少一区域；以及通过一投票/统计模块根据对应于文字特征的上述区域取得一车牌辨识结果。
附图说明
图1是显示根据本发明一实施例所述的车牌辨识系统的系统架构图。
图2是显示根据本发明一实施例所述的车牌辨识方法的流程图。
图3A是显示根据本发明一实施例所述的当前影像的示意图。
图3B是显示根据本发明一实施例所述的当前影像与历史背景影像的影像变化的示意图。
图4A～4D是显示根据本发明一些实施例所述的用以产生特征地图的经训练的矩阵的示意图。
图5A、5B是显示根据本发明一实施例所述的被判断为具有字元的区块的示意图。
图6是显示根据本发明一实施例所述的投票/统计模块的示意图。
图7A是显示根据本发明一实施例所述的当前影像的示意图。
图7B是显示根据本发明一实施例所述的车头影像的示意图。
图8是显示根据本发明一实施例所述的车牌文字区域的示意图。
图9是显示根据本发明另一实施例所述的车牌辨识方法的流程图。
附图标记说明：
100～车牌辨识系统
110～处理单元
120～储存单元
130～影像撷取单元
140～显示单元
801～804～对应于字元的区块
810～字元区块的联集
820～经扩张的车牌影像
S201～S205、S901～S906～步骤流程
具体实施方式
有关本发明的车牌辨识方法以及车牌辨识系统适用的其他范围将于接下来所提供的详述中清楚易见。必须了解的是下列的详述以及具体的实施例，当提出有关车牌辨识方法以及车牌辨识系统的示范实施例时，仅作为描述的目的以及并非用以限制本发明的范围。
图1是显示根据本发明一实施例所述的车牌辨识系统的系统架构图。车牌辨识系统100可实施于例如台式电脑、笔记本电脑或者平板电脑等的电子装置中，且车牌辨识系统100至少包含一处理单元110。处理单元110可通过多种方式实施，例如以专用硬件电路或者通用硬件(例如，单一处理器、具平行处理能力的多处理器、图形处理器或者其它具有运算能力的处理器)，且于执行与本发明各个模型以及流程有关的程序码或者软件时，提供之后所描述的功能。车牌辨识系统100还包括储存单元120，用以储存所取得的影像、执行过程中所需要的数据以及各式各样的电子文件，例如各种演算法和/或各个模型等。车牌辨识系统100更可包括影像撷取单元130，例如监视器、摄影机和/或相机等，用以取得至少一影像或者连续的视频影像，并将其回传至处理单元110。显示单元140可为显示面板(例如，薄膜液晶显示面板、有机发光二极管面板或者其它具显示能力的面板)，用以显示输入的字元、数字、符号、拖曳鼠标的移动轨迹或者应用程序所提供的使用者界面，以提供给使用者观看。车牌辨识系统100更可包括一输入装置(未显示)，例如滑鼠、触控笔或者键盘等，用以供使用者执行对应的操作。
请参阅图2。图2是显示根据本发明一实施例所述的车牌辨识系统的流程图。于步骤S201，影像撷取单元130取得一待处理影像。其中，为了加快影像处理的速度，当影像撷 取单元130为可撷取连续影像的监视器或者摄影机时，处理单元110更可通过事先将当前影像与历史影像进行比对以判断当前影像中是否有车辆或者其它物体进入拍摄范围中。举例来说，处理单元110可根据多个历史影像通过一前后景提取(Background Subtract)模块取得一历史背景影像，以供处理单元110快速地根据历史背景影像以及当前影像进行判断。当当前影像与历史背景影像之间的影像变化的面积或者影像变化量大于既定值时，则处理单元110判断有车辆或者其它物体进入当前影像所对应的拍摄范围中，接着再对当前影像执行后续的动作。举例来说，图3A为一当前影像的示意图，而图3B则为当前影像与历史背景影像之间的影像变化的示意图。其中，根据图3B的内容可得知影像变化的面积约为37％，若既定值设定为35％，则处理单元110即可判断该当前影像中出现车辆或者其它物体。
于步骤S202，处理单元110接收到待处理影像，并通过一特征地图提取模块取得多个特征地图。其中，特征地图提取模块可通过强化字元特征的矩阵进行训练，其主要是用以凸显影像中的英文字母或者数字等字元。图4A～4D是显示根据本发明一些实施例所述的用以取得特征地图的多个经训练的矩阵的示意图。于步骤S203，于取得特征地图后，处理单元110通过一字元辨识模型根据特征地图撷取对应于每个字元的区块以及对应的坐标。其中，字元辨识模块为基于神经网络的模块，其主要是以对应至各种不同字母(即A～Z)、数字(即0～9)的多个影像作为训练数据，以准确地辨识出影像中每个字元的区块以及其位置。举例来说，图5A中所显示的多个区块为被判断为具有字元的区块。其中，字元辨识模块可直接找出每个字元所对应的区块，而不需要事先对车牌中的字元进行分割。然而，如图5A所示，由于仍有一些不具有字元的区块(例如周遭环境类似于字元的影像或者影像中的噪声等)被误判，因此处理单元110更通过选择可信度较高的区域(例如通过选取具有较多重叠的区域)作为具有字元的区块，并同时撷取区域的坐标，以作为车牌字元排序的根据。举例来说，如图5B所示，粗框线所选取的区域即为最后所撷取的具有字元的区域，而通过上述的方法即可有效地过滤被误判的区块。
接着，进入步骤S204，处理单元110于取得所有字元以及其所对应的坐标后，即根据每个字元以及坐标的排序取得一车牌辨识结果。根据本发明一实施例，处理单元110更可通过一投票/统计模块对多个车牌影像进行投票以提高车牌辨识结果的准确度。于取得车牌的各个字元以及排列顺序后，处理单元110可通过车牌分群规则将车牌区分类为至少两个分群。其中，车牌分群规则可包括一车牌命名分群规则、一英文字区与数字字区分群规则、一破折号分群规则以及一字元相对位置分群规则等。将车牌进行分群后，处理单元 110接着对每个分群的辨识结果进行投票，当每个分群中皆出现投票分数大于门槛值的辨识结果时，即可产生一最终车牌辨识结果。举例来说，图6是显示根据本发明一实施例所述的投票/统计模块的示意图。于此实施例中，处理单元110是采用破折号分群规则将车牌以破折号的位置为基准分为两个分群(即破折号的左半部分以及右半部分)。接着，于取得两个分群的辨识结果后，对每个分群中不同的辨识结果进行投票，若出现重复的辨识结果则累加投票分数。举例来说，如图6所示，”2806”于第五组辨识结果出现第一次重复，则”2806”的投票分数累加为2，而”J3”则于第四组辨识结果出现第一次重复，则”J3”的投票分数则于第四组时即累加为2，以此类推。假设投票分数的门槛值已预先设定为4，则破折号右半部分的分群则在第六组辨识结果即可确定为”J3”，而破折号左半部分的分群则在第七组辨识结果才能确定为”2806”。接着，处理单元110于完成第七组的车牌辨识结果后，即可输出”2806-J3”的最终车牌辨识结果。
根据本发明另一实施例，处理单元110更可根据辨识结果的时间排序赋予辨识结果不同的权重。举例来说，较新的辨识结果赋予较大的权重，而较旧的辨识结果则赋予较小的权重，藉此以加快最终车牌辨识结果的收敛速度。
此外，根据本发明另一实施例，为了加速处理单元110对车牌信息的处理速度，于取得具有车牌影像的当前影像后，处理单元110更可通过一车头影像撷取模块或者一车尾影像撷取模块于当前影像中取得一车头影像或者一车尾影像，以缩小欲处理的影像面积。其中，车头影像撷取模块或者车尾影像撷取模块是通过多个图像特征(例如Haar Feature、HOG、LBP等)搭配分类器(Cascade Classifier、Ada boost或者SVM)来训练各种车头影像或者车尾影像，以从当前影像中取得车头影像或者车尾影像。举例来说，图7A是显示一当前影像的示意图，而图7B则显示通过车头影像撷取模块所取得的车头影像的示意图。
根据本发明另一实施例，于取得车头影像或者车尾影像后，为了更进一步地缩小处理单元110欲处理的影像区域大小，处理单元110更可通过一车牌字元区域检测模型自车头影像或者车尾影像中取得车牌附近的区域。其中，车牌字元区域检测模型亦通过多个图像特征(例如Haar Feature、HOG、LBP等)搭配分类器(Cascade Classifier、Ada boost或者SVM)来训练各个字元影像，以从车头影像或者车尾影像中找出各个字元。举例来说，如图8中所示，处理单元110是通过车牌字元区域检测模型是自车头影像中找出四个具有字元的区域801～804。接着，处理单元110将区域801～804联集以取得另一更大的区域810，并根据车牌的格式放大区域810进行扩张，以取得包含所有车牌字元的另一区域。举例来 说，车牌是具有六个字元，而车牌字元区域检测模型仅自车头影像中找出四个具有字元的区域，因此为了确保待处理影像中包含车牌中的所有字元，处理单元110更可根据被找到的字元数目决定向外扩张的倍率。举例来说，如图8所示，由于处理单元110已找到四个字元，因此处理单元110从区域810的四个边向外扩张约一倍(如区域820所示)，如此即可确保所有的字元皆被包括于待处理影像中。换言之，若处理单元110仅找到一个具有字元的区域，则处理单元110适应性地增加扩张的倍率(例如向左侧以及右侧扩张十倍)，以确保所有的字元皆被包括于待处理影像中。其中，扩张的倍率可根据使用者的需求进行调整，前述的实施例仅用以作为说明的用途，本发明并不以此为限。相较于车头影像或者车尾影像，通过车牌字元区域检测模型所取得的待处理影像更精准地缩小影像的面积，以更进一步地提升运算的速度。值得注意的是，由于车牌字元区域检测模型的主要功能仅用以找出可能具有字元的区域，并非用以精确地辨识字元，因此相较于字元辨识模型，车牌字元区域检测模型是属于弱分类器，即其检测准确率较低但计算速度较快。此外，车牌字元区域检测模型与车头影像撷取模块或者车尾影像撷取模块是使用不同的图像特征以及分类器。
于步骤S205，为了进一步地提高字元辨识模型的准确率，处理单元110更将每个影像以及对应的辨识结果作为训练数据以更新字元辨识模型。其中，上述辨识结果包含正确的车牌辨识结果以及不正确的车牌辨识结果，藉此以降低字元辨识模块的辨识误差，并可间接地加快车牌辨识系统的处理速度。
图9是显示根据本发明另一实施例所述的车牌辨识方法的流程图。于步骤S901，影像撷取单元130取得至少一待处理影像。于步骤S902，处理单元110自影像撷取单元130接收所取得的待处理影像，并通过特征地图提取模块取得多个特征地图。其中，特征地图中所包含的信息包括对应于不同空间频率(例如自低频至高频)的多个目标特征，而目标特征可包含车牌上的字元特征、车牌外型特征、背景特征、车辆信息特征(例如方向镜、车型、车轮等代表汽车的特征)。此外，特征地图提取模块可通过包含前述目标特征的矩阵进行训练。于步骤S903，于取得多个特征地图后，处理单元110根据特征地图通过一目标位置提取模块找出具有前述目标特征的区域。
其中，根据本发明一实施例，处理单元110可通过类聚的方式或者自定义的尺寸于特征地图上每个既定像素提取一个框，并根据特征地图提取模块判断每个框中可能包含的特征，并给予每个框对应于每个目标特征类型的分数。或者，根据本发明另一实施例，处理单元110先通过一简易分类器取得特征地图的目标敏感分数图，即于特征地图上找出具有 目标特征的多个目标特征点或者目标特征区域，接着利用具有不同大小的框圈选出位于该目标特征点附近的多个区域，并给予该些区域对应于每个目标特征的分数。
接着，处理单元110于取得对应于每个框的每个目标特征类型的所有分数后，进入步骤S904，处理单元110通过一目标候选分类模块以非极大值抑制的方式仅保留具有最高分数且分数大于既定值的目标特征，以对每个框所对应的位置进行分类。举例来说，某一个框对应于背景特征的分数最大，且大于既定值，则处理单元110将该框分类为对应于背景特征的框。此外，当某一特定框对应于每个目标特征的分数皆未大于既定值时，将该区域分类为非目标特征。此外，处理单元110更可通过目标候选分类模块将具有相同目标特征且彼此相邻的多个框集结为较大的区域，以利于后续的辨识流程。接着，处理单元110仅保留对应至字元特征的区域，并进入步骤S905。于步骤S905，处理单元110根据每个字元以及坐标的排序(例如由左至右、由上自下)取得一车牌辨识结果。如前所述，处理单元110可通过前述的投票/统计模块对多个车牌影像进行投票以提高车牌辨识结果的准确度。其中，步骤S905所述的车牌辨识方式是与步骤S204的车牌辨识方法类似，在此即不加以描述以精简说明。
最后，于步骤S906，处理单元110更将每个待处理影像以及对应的辨识结果作为训练数据以更新字元辨识模型。其中，上述辨识结果包含正确的车牌辨识结果以及不正确的车牌辨识结果，藉此以降低字元辨识模块的辨识误差。
综上所述，根据本发明一些实施例所提出的车牌辨识方法以及车牌辨识系统，通过前述的车牌影像撷取步骤以及车牌字元辨识步骤，在视角不佳或者变化复杂的环境下，仍可维持快速的辨识速度以及高准确率，且通过不断地将辨识结果作为训练数据，将可更进一步地降低车牌辨识的误差并间接地加快车牌辨识系统的计算速度。
以上叙述许多实施例的特征，使所属技术领域技术人员能够清楚理解本说明书的形态。所属技术领域技术人员能够理解其可利用本发明公开内容为基础以设计或变动其他制程及结构而完成相同于上述实施例的目的及/或达到相同于上述实施例的优点。所属技术领域技术人员亦能够理解不脱离本发明的精神和范围的等效构造可在不脱离本发明的精神和范围内作任意的变动、替代与润饰。

Claims (25)

  1. 一种车牌辨识方法，包括以下步骤：
    取得包含所有车牌字元的一待处理影像；
    通过一特征地图提取模块提取具有上述待处理影像的字元特征的多个特征地图；
    通过基于神经网络的一字元辨识模型根据上述特征地图撷取对应于每个字元的区块以及坐标；以及
    根据每个字元的上述区块以及上述坐标取得一车牌辨识结果。
  2. 如权利要求1所述的车牌辨识方法，还包括以下步骤：
    接收一原始影像；
    比较上述原始影像与一历史背景影像以取得影像变化量；以及
    判断上述影像变化量是否大于一预设值；
    其中，当上述影像变化量大于上述预设值时，产生包含所有车牌字元的上述待处理影像。
  3. 如权利要求2所述的车牌辨识方法，还包括一步骤：
    通过一前后景提取模块提取上述历史背景影像。
  4. 如权利要求1所述的车牌辨识方法，还包括以下步骤：
    接收一原始影像；
    通过一车头影像撷取模块或者一车尾影像撷取模块利用第一图像特征以及一第一分类器自上述原始影像中取得一车头影像或者一车尾影像；以及
    通过一车牌字元区域检测模型根据上述车头影像或者上述车尾影像取得包含所有车牌字元的上述待处理影像。
  5. 如权利要求4所述的车牌辨识方法，还包括以下步骤：
    通过上述车牌字元区域检测模型利用第二图像特征以及一第二分类器于上述车头影像或者上述车尾影像中取得至少一字元区块；
    根据上述字元区块的数量决定一放大倍率；以及
    基于上述字元区块根据上述放大倍率取得上述待处理影像。
  6. 如权利要求1所述的车牌辨识方法，还包括以下步骤：
    接收多个上述车牌辨识结果；
    根据一车牌分群规则将每个上述车牌辨识结果分类为至少两个分群；
    对每个上述分群中的每个子辨识结果进行投票；以及
    当每个上述分群中的任一上述子辨识结果的一投票分数大于一门槛值时，根据上述子辨识结果产生一最终车牌辨识结果。
  7. 如权利要求6所述的车牌辨识方法，还包括以下步骤：
    根据所有上述车牌辨识结果的时序排列赋予每个上述车牌辨识结果一权重；以及
    当每个上述分群中的任一上述子辨识结果的一权重总和大于上述门槛值时，根据上述子辨识结果产生上述最终车牌辨识结果。
  8. 如权利要求6所述的车牌辨识方法，其中上述车牌分群规则包括一车牌命名分群规则、一英文字区与数字字区分群规则、一破折号分群规则以及一字元相对位置分群规则。
  9. 如权利要求6项或第7所述的车牌辨识方法，还包括一步骤：
    根据上述车牌辨识结果和/或上述最终车牌辨识结果更新上述字元辨识模型。
  10. 一种车牌辨识系统，包括：
    一影像撷取单元，用以撷取至少一原始影像；以及
    一处理单元，用以：
    自上述影像撷取单元接收上述原始影像；
    根据上述原始影像取得包含所有车牌字元的一待处理影像；
    通过一特征地图提取模块提取具有上述待处理影像的字元特征的多个特征地图；
    通过基于神经网络的一字元辨识模型根据上述特征地图撷取对应于每个字元的区块以及坐标；以及
    根据每个字元的上述区块以及上述坐标取得一车牌辨识结果。
  11. 如权利要求10所述的车牌辨识系统，其中上述处理单元更用以：
    比较上述原始影像与一历史背景影像以取得影像变化量；以及
    判断上述影像变化量是否大于一预设值；
    其中，当上述影像变化量大于上述预设值时，上述处理单元产生包含所有车牌字元的上述待处理影像。
  12. 如权利要求11所述的车牌辨识系统，其中上述处理单元更通过一前后景提取模块提取上述历史背景影像。
  13. 如权利要求10所述的车牌辨识系统，其中上述处理单元更用以：
    通过一车头影像撷取模块或者一车尾影像撷取模块利用第一图像特征以及一第一分类器自上述原始影像中取得一车头影像或者一车尾影像；以及
    通过一车牌字元区域检测模型根据上述车头影像或者上述车尾影像取得包含所有车牌字元的上述待处理影像。
  14. 如权利要求13所述的车牌辨识系统，其中上述处理单元更用以：
    通过上述车牌字元区域检测模型利用第二图像特征以及一第二分类器于上述车头影像或者上述车尾影像中取得至少一字元区块；
    根据上述字元区块的数量决定一放大倍率；以及
    基于上述字元区块根据上述放大倍率取得上述待处理影像。
  15. 如权利要求10所述的车牌辨识系统，其中上述处理单元更用以：
    接收多个上述车牌辨识结果；
    根据一车牌分群规则将每个上述车牌辨识结果分类为至少两个分群；
    对每个上述分群中的每个子辨识结果进行投票；以及
    当每个上述分群中的任一上述子辨识结果的一投票分数大于一门槛值时，根据上述子辨识结果产生一最终车牌辨识结果。
  16. 如权利要求15所述的车牌辨识系统，其中上述处理器更用以：
    根据所有上述车牌辨识结果的时序排列赋予每个上述车牌辨识结果一权重；以及
    当每个上述分群中的任一上述子辨识结果的一权重总和大于上述门槛值时，根据上述子辨识结果产生上述最终车牌辨识结果。
  17. 如权利要求15所述的车牌辨识系统，其中上述车牌分群规则包括一车牌命名分群规则、一英文字区与数字字区分群规则、一破折号分群规则以及一字元相对位置分群规则。
  18. 如权利要求15项或第16所述的车牌辨识系统，其中上述处理器更用以：
    根据上述车牌辨识结果和/或上述最终车牌辨识结果更新上述字元辨识模型。
  19. 一种车牌辨识方法，包括以下步骤：
    取得一待处理影像；
    通过一特征地图提取模块取得具有多个目标特征的多个特征地图；
    通过一目标位置提取模块取得每个上述特征地图中具有上述目标特征的至少一区域，并给予每个上述特征地图的每个上述框对应于每个上述目标特征的分数；
    通过一目标候选分类模块根据上述分数对每个上述特征地图的每个上述框进行分类，并保留对应于文字特征的至少一区域；以及
    通过一投票/统计模块根据对应于上述文字特征的上述区域取得一车牌辨识结果。
  20. 如权利要求19所述的车牌辨识方法，其中上述目标特征包括对应于不同空间频率的上述字元特征、车牌外型特征、背景特征或者车辆特征。
  21. 如权利要求19所述的车牌辨识方法，还包括：
    通过类聚的方式或者自定义尺寸于上述特征地图上每隔既定像素提取上述框，以通过上述目标位置提取模块给予每个上述特征地图的每个上述框对应于每个上述目标特征的上述分数。
  22. 如权利要求19所述的车牌辨识方法，还包括：
    通过一简易分类器取得上述特征地图的至少一目标特征点，利用具有不同大小的多个框圈选出位于上述目标特征点附近的多个区域，以及通过上述特征地图提取模块给予每个上述特征地图的每个上述框对应于每个上述目标特征的上述分数。
  23. 如权利要求19所述的车牌辨识方法，还包括：
    通过上述目标候选分类模块以一非极大值抑制的方式保留具有最高分数且分数大于一既定值的上述文字特征。
  24. 如权利要求19所述的车牌辨识方法，还包括：
    接收多个上述车牌辨识结果；
    通过上述投票/统计模块根据一车牌分群规则将每个上述车牌辨识结果分类为至少两个分群；
    通过上述投票/统计模块对每个上述分群中的每个子辨识结果进行投票；以及
    当每个上述分群中的任一上述子辨识结果的一投票分数大于一门槛值时，上述投票/统计模块根据上述子辨识结果产生一最终车牌辨识结果。
  25. 如权利要求19所述的车牌辨识方法，还包括：
    根据上述车牌辨识结果更新上述特征地图提取模块。
PCT/CN2019/072542 2018-03-14 2019-01-21 车牌辨识方法以及其系统 Ceased WO2019174405A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
US16/980,747 US11443535B2 (en) 2018-03-14 2019-01-21 License plate identification method and system thereof
JP2020548723A JP7044898B2 (ja) 2018-03-14 2019-01-21 ナンバープレート認識方法、および、そのシステム
EP19768532.4A EP3767530A4 (en) 2018-03-14 2019-01-21 PROCEDURE FOR IDENTIFICATION OF LICENSE PLATES AND THE SYSTEM THEREFORE
US17/881,218 US20220375236A1 (en) 2018-03-14 2022-08-04 License plate identification method and system thereof

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810208466.9A CN110276342B (zh) 2018-03-14 2018-03-14 车牌辨识方法以及其系统
CN201810208466.9 2018-03-14

Related Child Applications (2)

Application Number Title Priority Date Filing Date
US16/980,747 A-371-Of-International US11443535B2 (en) 2018-03-14 2019-01-21 License plate identification method and system thereof
US17/881,218 Division US20220375236A1 (en) 2018-03-14 2022-08-04 License plate identification method and system thereof

Publications (1)

Publication Number Publication Date
WO2019174405A1 true WO2019174405A1 (zh) 2019-09-19

Family

ID=67907354

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/072542 Ceased WO2019174405A1 (zh) 2018-03-14 2019-01-21 车牌辨识方法以及其系统

Country Status (5)

Country Link
US (2) US11443535B2 (zh)
EP (1) EP3767530A4 (zh)
JP (1) JP7044898B2 (zh)
CN (1) CN110276342B (zh)
WO (1) WO2019174405A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111079744A (zh) * 2019-12-06 2020-04-28 鲁东大学 适用于复杂光照环境的车辆车牌智能识别方法及装置
CN111340045A (zh) * 2020-02-12 2020-06-26 浙江大华技术股份有限公司 车牌号码的识别方法、装置及存储介质
CN111582261A (zh) * 2020-04-30 2020-08-25 浙江大华技术股份有限公司 一种非机动车的车牌识别方法和车牌识别装置
CN112560608A (zh) * 2020-12-05 2021-03-26 江苏爱科赛尔云数据科技有限公司 一种车辆车牌识别方法
JP2021082056A (ja) * 2019-11-20 2021-05-27 日本放送協会 文字領域検出モデル学習装置およびそのプログラム、ならびに、文字領域検出装置およびそのプログラム
JP2022504404A (ja) * 2019-09-27 2022-01-13 シェンチェン センスタイム テクノロジー カンパニー リミテッド テキストシーケンス認識方法及びその装置、電子機器並びに記憶媒体

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110276342B (zh) * 2018-03-14 2023-04-18 台达电子工业股份有限公司 车牌辨识方法以及其系统
US11567632B2 (en) 2018-07-03 2023-01-31 Apple Inc. Systems and methods for exploring a geographic region
US11275934B2 (en) * 2019-11-20 2022-03-15 Sap Se Positional embeddings for document processing
US20210356288A1 (en) 2020-05-15 2021-11-18 Apple Inc. User interfaces for providing navigation directions
US11846515B2 (en) 2020-06-11 2023-12-19 Apple Inc. User interfaces for customized navigation routes
US12281912B2 (en) 2021-06-07 2025-04-22 Apple Inc. User interfaces for maps and navigation
CN119223321A (zh) 2021-06-07 2024-12-31 苹果公司 用于地图和导航的用户界面
CN116095459B (zh) * 2021-11-02 2025-05-30 睿能创意公司 影像撷取方法
CN116095458B (zh) * 2021-11-02 2025-02-11 睿能创意公司 影像撷取方法
EP4235595B1 (en) * 2022-02-28 2025-10-29 Leica Biosystems Imaging, Inc. High and low frequency feature map generation for h&e pathology images
CN114821019B (zh) * 2022-03-29 2025-02-25 烽火通信科技股份有限公司 车牌号码识别方法、装置、设备及可读存储介质
CN114998883B (zh) * 2022-05-30 2024-10-18 阿波罗智联(北京)科技有限公司 车牌识别方法、装置、电子设备、智能交通设备
CN115359460B (zh) * 2022-10-20 2023-02-17 小米汽车科技有限公司 用于车辆的图像识别方法、装置、车辆和存储介质
CN116580461B (zh) * 2023-05-12 2025-12-09 中国工商银行股份有限公司 签名验证方法、装置、计算机设备和存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106096607A (zh) * 2016-06-12 2016-11-09 湘潭大学 一种车牌识别方法
CN106980854A (zh) * 2017-03-29 2017-07-25 珠海习悦信息技术有限公司 车牌号码识别方法、装置、存储介质及处理器
CN107038442A (zh) * 2017-03-27 2017-08-11 新智认知数据服务有限公司 一种基于深度学习的车牌检测和整体识别方法
CN107239778A (zh) * 2017-06-09 2017-10-10 中国科学技术大学 高效准确的车牌识别方法
CN107704857A (zh) * 2017-09-25 2018-02-16 北京邮电大学 一种端到端的轻量级车牌识别方法及装置
CN107784303A (zh) * 2016-12-15 2018-03-09 平安科技（深圳）有限公司 车牌识别方法及装置

Family Cites Families (27)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11306283A (ja) 1998-04-24 1999-11-05 Chuo Spring Co Ltd ナンバープレート情報読取装置
JP4174279B2 (ja) 2002-09-19 2008-10-29 日本放送協会 映像オブジェクト識別・追跡装置、その方法及びそのプログラム
TWI236638B (en) 2004-03-24 2005-07-21 Univ Chung Yuan Christian Multiple recognition system and method for license plates
WO2008025092A1 (en) * 2006-09-01 2008-03-06 Sensen Networks Pty Ltd Method and system of identifying one or more features represented in a plurality of sensor acquired data sets
JP4901676B2 (ja) 2007-09-28 2012-03-21 株式会社東芝 ナンバープレート情報処理装置及びナンバープレート情報処理方法
CN101303803A (zh) 2008-06-11 2008-11-12 北京中星微电子有限公司 一种车牌识别的方法和系统
TWI431559B (zh) * 2010-10-29 2014-03-21 Univ Nat Chiao Tung 車牌偵測辨識方法及其系統
US8447112B2 (en) * 2010-12-17 2013-05-21 Xerox Corporation Method for automatic license plate recognition using adaptive feature set
US8666117B2 (en) * 2012-04-06 2014-03-04 Xerox Corporation Video-based system and method for detecting exclusion zone infractions
US9014432B2 (en) * 2012-05-04 2015-04-21 Xerox Corporation License plate character segmentation using likelihood maximization
CN103366577B (zh) * 2013-05-16 2015-12-23 王斌 车辆牌证真伪识别系统
CN103679191B (zh) * 2013-09-04 2017-02-22 西交利物浦大学 基于静态图片的自动套牌车检测方法
NL2012327B1 (en) * 2013-12-13 2016-06-21 Utc Fire & Security B V Selective intrusion detection systems.
JP6471934B2 (ja) * 2014-06-12 2019-02-20 パナソニックＩｐマネジメント株式会社 画像認識方法、カメラシステム
JP2016053763A (ja) 2014-09-02 2016-04-14 キヤノン株式会社 画像処理装置、画像処理方法及びプログラム
CN104239867B (zh) 2014-09-17 2017-12-01 深圳市捷顺科技实业股份有限公司 车牌定位方法及系统
CN106599900B (zh) * 2015-10-20 2020-04-21 华中科技大学 一种识别图像中的字符串的方法和装置
CN106469299B (zh) * 2016-08-31 2019-07-19 北京邮电大学 一种车辆搜索方法及装置
CN106529532A (zh) * 2016-11-07 2017-03-22 湖南源信光电科技有限公司 一种基于积分特征通道与灰度投影的车牌识别系统
CN106407981B (zh) * 2016-11-24 2019-04-23 北京文安智能技术股份有限公司 一种车牌识别方法、装置及系统
CN108121984B (zh) 2016-11-30 2021-09-21 杭州海康威视数字技术股份有限公司 一种字符识别方法及装置
CN106650740B (zh) * 2016-12-15 2019-08-13 深圳市华尊科技股份有限公司 一种车牌识别方法及终端
WO2018112441A1 (en) * 2016-12-15 2018-06-21 Progenity Inc. Ingestible device and associated methods
CN106845487B (zh) * 2016-12-30 2020-04-17 佳都新太科技股份有限公司 一种端到端的车牌识别方法
CN107220638A (zh) * 2017-07-03 2017-09-29 深圳市唯特视科技有限公司 一种基于深度学习卷积神经网络的车牌检测识别方法
CN110276342B (zh) * 2018-03-14 2023-04-18 台达电子工业股份有限公司 车牌辨识方法以及其系统
CN111524365B (zh) * 2020-05-11 2021-12-17 西安电子科技大学 一种利用多地磁传感器进行车型分类的方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106096607A (zh) * 2016-06-12 2016-11-09 湘潭大学 一种车牌识别方法
CN107784303A (zh) * 2016-12-15 2018-03-09 平安科技（深圳）有限公司 车牌识别方法及装置
CN107038442A (zh) * 2017-03-27 2017-08-11 新智认知数据服务有限公司 一种基于深度学习的车牌检测和整体识别方法
CN106980854A (zh) * 2017-03-29 2017-07-25 珠海习悦信息技术有限公司 车牌号码识别方法、装置、存储介质及处理器
CN107239778A (zh) * 2017-06-09 2017-10-10 中国科学技术大学 高效准确的车牌识别方法
CN107704857A (zh) * 2017-09-25 2018-02-16 北京邮电大学 一种端到端的轻量级车牌识别方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3767530A4 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2022504404A (ja) * 2019-09-27 2022-01-13 シェンチェン センスタイム テクノロジー カンパニー リミテッド テキストシーケンス認識方法及びその装置、電子機器並びに記憶媒体
JP7123255B2 (ja) 2019-09-27 2022-08-22 シェンチェン センスタイム テクノロジー カンパニー リミテッド テキストシーケンス認識方法及びその装置、電子機器並びに記憶媒体
JP2021082056A (ja) * 2019-11-20 2021-05-27 日本放送協会 文字領域検出モデル学習装置およびそのプログラム、ならびに、文字領域検出装置およびそのプログラム
JP7373367B2 (ja) 2019-11-20 2023-11-02 日本放送協会 文字領域検出モデル学習装置およびそのプログラム、ならびに、文字領域検出装置およびそのプログラム
CN111079744A (zh) * 2019-12-06 2020-04-28 鲁东大学 适用于复杂光照环境的车辆车牌智能识别方法及装置
CN111079744B (zh) * 2019-12-06 2020-09-01 鲁东大学 适用于复杂光照环境的车辆车牌智能识别方法及装置
CN111340045A (zh) * 2020-02-12 2020-06-26 浙江大华技术股份有限公司 车牌号码的识别方法、装置及存储介质
CN111340045B (zh) * 2020-02-12 2023-09-01 浙江大华技术股份有限公司 车牌号码的识别方法、装置及存储介质
CN111582261A (zh) * 2020-04-30 2020-08-25 浙江大华技术股份有限公司 一种非机动车的车牌识别方法和车牌识别装置
CN111582261B (zh) * 2020-04-30 2024-01-19 浙江大华技术股份有限公司 一种非机动车的车牌识别方法和车牌识别装置
CN112560608A (zh) * 2020-12-05 2021-03-26 江苏爱科赛尔云数据科技有限公司 一种车辆车牌识别方法
CN112560608B (zh) * 2020-12-05 2024-05-24 江苏爱科赛尔云数据科技有限公司 一种车辆车牌识别方法

Also Published As

Publication number Publication date
CN110276342A (zh) 2019-09-24
US20210004627A1 (en) 2021-01-07
US11443535B2 (en) 2022-09-13
CN110276342B (zh) 2023-04-18
EP3767530A4 (en) 2021-05-19
JP2021518944A (ja) 2021-08-05
JP7044898B2 (ja) 2022-03-30
US20220375236A1 (en) 2022-11-24
EP3767530A1 (en) 2021-01-20

Similar Documents

Publication Publication Date Title
US20220375236A1 (en) License plate identification method and system thereof
US11164027B2 (en) Deep learning based license plate identification method, device, equipment, and storage medium
CN108121986B (zh) 目标检测方法及装置、计算机装置和计算机可读存储介质
US8792722B2 (en) Hand gesture detection
US8750573B2 (en) Hand gesture detection
JP6397986B2 (ja) 画像オブジェクト領域の認識方法及び装置
CN115461792B (zh) 手写文本识别方法、装置和系统，手写文本搜索方法和系统，以及计算机可读存储介质
WO2023284502A1 (zh) 图像处理方法、装置、设备和存储介质
CN113033543B (zh) 曲形文本识别方法、装置、设备及介质
WO2021051545A1 (zh) 基于行为识别模型的摔倒动作判定方法、装置、计算机设备及存储介质
US10242294B2 (en) Target object classification using three-dimensional geometric filtering
CN111488732B (zh) 一种变形关键词检测方法、系统及相关设备
US11250269B2 (en) Recognition method and apparatus for false detection of an abandoned object and image processing device
CN114627534B (zh) 活体判别方法及电子设备、存储介质
CN110598703A (zh) 一种基于深度神经网络的ocr识别方法及装置
CN118522019B (zh) 文本识别方法、电子设备以及存储介质
US20150139547A1 (en) Feature calculation device and method and computer program product
CN116543261B (zh) 用于图像识别的模型训练方法、图像识别方法设备及介质
TWI690857B (zh) 車牌辨識方法以及其系統
Moumen et al. Real-time Arabic scene text detection using fully convolutional neural networks
CN113569838A (zh) 基于文本检测算法的文本识别方法及装置
Kwon et al. Novel rifle number recognition based on improved YOLO in military environment
US20130279808A1 (en) Complex-object detection using a cascade of classifiers
CN119887681A (zh) 基于视觉-语言提示的工业缺陷检测方法
CN107545261A (zh) 文本检测的方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19768532

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2020548723

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2019768532

Country of ref document: EP

Effective date: 20201014

WWW Wipo information: withdrawn in national office

Ref document number: 2019768532

Country of ref document: EP