WO2020186790A1 - 车型识别方法、装置、计算机设备及存储介质 - Google Patents

车型识别方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2020186790A1
WO2020186790A1 PCT/CN2019/118435 CN2019118435W WO2020186790A1 WO 2020186790 A1 WO2020186790 A1 WO 2020186790A1 CN 2019118435 W CN2019118435 W CN 2019118435W WO 2020186790 A1 WO2020186790 A1 WO 2020186790A1
Authority
WO
WIPO (PCT)
Prior art keywords
vehicle
frame number
image
identified
candidate area
Prior art date
Application number
PCT/CN2019/118435
Other languages
English (en)
French (fr)
Inventor
李晨光
Original Assignee
平安科技(深圳)有限公司
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 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2020186790A1 publication Critical patent/WO2020186790A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/28Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/44Local 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

Definitions

  • This application relates to the field of image recognition technology, and in particular to a vehicle type recognition method, device, computer equipment and storage medium.
  • the embodiments of the present application provide a method, device, computer equipment, and storage medium for vehicle identification, aiming to solve the problem of manually inputting relevant vehicle type details when registering a vehicle type in the prior art, which is not only time-consuming, but also prone to human factors. Lead to the problem of inaccurate vehicle information registration.
  • an embodiment of the present application provides a vehicle model identification method, which includes:
  • an image of a vehicle to be identified is detected and the image of the vehicle to be identified includes vehicle interior parts, obtaining the vehicle model according to the frame number in the image of the vehicle to be identified;
  • the unique model information corresponding to the image of the vehicle to be identified is acquired.
  • an embodiment of the present application provides a vehicle type identification device, which includes:
  • a vehicle model acquisition unit configured to acquire the vehicle model according to the frame number in the vehicle image to be identified if the image of the vehicle to be identified is detected and the image of the vehicle to be identified includes vehicle interior parts;
  • the input sequence acquisition unit is used to identify and acquire each vehicle interior part in the picture of the vehicle to be identified to obtain a corresponding vehicle interior part sequence
  • a configuration level acquisition unit configured to use the vehicle interior part sequence corresponding to the vehicle picture to be recognized as the input of the pre-trained convolutional neural network model to obtain the vehicle configuration level corresponding to the vehicle picture to be recognized;
  • the vehicle type identification unit is configured to obtain unique vehicle type information corresponding to the vehicle image to be identified according to the vehicle model and vehicle configuration level corresponding to the vehicle image to be identified.
  • an embodiment of the present application provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and running on the processor, and the processor executes the computer
  • the program implements the vehicle identification method described in the first aspect.
  • the embodiments of the present application also provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the above-mentioned The vehicle identification method described on the one hand.
  • FIG. 1 is a schematic diagram of an application scenario of a vehicle type identification method provided by an embodiment of the application
  • FIG. 2 is a schematic flowchart of a method for identifying a vehicle model provided by an embodiment of the application
  • FIG. 3 is a schematic diagram of another flow chart of the vehicle type identification method provided by an embodiment of the application.
  • FIG. 4 is a schematic diagram of a sub-process of the vehicle type identification method provided by an embodiment of the application.
  • FIG. 5 is a schematic diagram of another sub-process of the vehicle type identification method provided by an embodiment of the application.
  • FIG. 6 is a schematic block diagram of a vehicle type identification device provided by an embodiment of the application.
  • FIG. 7 is another schematic block diagram of the vehicle type identification device provided by an embodiment of the application.
  • FIG. 8 is a schematic block diagram of a subunit of the vehicle type recognition device provided by an embodiment of the application.
  • FIG. 9 is a schematic block diagram of another subunit of the vehicle type recognition device provided by an embodiment of the application.
  • FIG. 10 is a schematic block diagram of a computer device provided by an embodiment of the application.
  • FIG. 1 is a schematic diagram of an application scenario of the vehicle type identification method provided by an embodiment of the application
  • FIG. 2 is a schematic flowchart of the vehicle type identification method provided by an embodiment of the application.
  • the vehicle type identification method is applied in a server, The method is executed by application software installed in the server.
  • the method includes steps S110 to S140.
  • the uploader such as smart phones, tablet computers, and other smart terminals
  • the uploader takes pictures of the vehicle from multiple perspectives (especially pictures of the interior of the vehicle)
  • the server receives and detects the image of the vehicle to be identified, and the image of the vehicle to be identified includes vehicle interior parts, obtain the vehicle model according to the frame number in the image of the vehicle to be identified. Through the identification of the frame number to determine the vehicle model, the value estimation information of the vehicle can be quickly obtained. If the vehicle needs to be accurately estimated, the vehicle interior components need to be further identified.
  • the frame number also known as Vehicle Identification Number (Vehicle Identification Number, abbreviated as VIN)
  • VIN Vehicle Identification Number
  • the frame number is a set of seventeen alphanumeric numbers, a set of unique numbers used on cars, and can identify the manufacturer of the car , Engine, chassis serial number and other performance information. To avoid confusion with the number 1,0, the English letters "I”, "O", and "Q” will not be used.
  • the VIN code of a car can mostly be found on the left side of the dashboard and under the windshield, usually written on a nameplate. Therefore, the process of recognizing the frame number in the picture of the vehicle to be recognized is similar to the process of license plate recognition.
  • the method before step S110, the method further includes:
  • S102 Label each vehicle interior part in the vehicle picture as the input of the to-be-trained convolutional neural network, and use the vehicle configuration level corresponding to the vehicle picture as the output of the to-be-trained convolutional neural network.
  • the convolutional neural network is trained to obtain a convolutional neural network model for identifying vehicle configuration levels.
  • a crawler tool is used to obtain a list of preset URLs (the list of URLs stores a large number of vehicle images including vehicle interior details), and at this time, the interior components included in the crawled vehicle images can be Make an annotation.
  • the main concern is: judging whether the steering wheel is multi-functional, judging whether there is a sunroof, judging whether the transmission is manual or automatic, judging whether the leather seat can be determined by the seat picture Judging by the reflectance of the seat part, the leather is generally not reflective, and the central part of the photo taken from the non-leather seat is obviously reflective, and there will be a white flower), and judge whether there is an air-conditioning outlet in the rear.
  • it is marked as 1 when it is a multifunctional steering wheel, and it is marked as 0 when it is not a multifunctional steering wheel; it is marked as 1 when there is a sunroof, and it is marked as 0 when there is no sunroof; it is marked as 1 when it is an automatic transmission, and it is marked as a manual transmission.
  • Configuration indicates that the vehicle configuration level is low-end configuration.
  • a vehicle is labeled as [1,1,1,1], and its corresponding vehicle configuration level is 1 to indicate a high-end configuration car; for example, a vehicle is labeled as [0,0,0,0 ,0], its corresponding vehicle configuration level is 3 to indicate a low-end configuration car; for example, a vehicle is labeled as [0,1,0,1,0], and its corresponding vehicle configuration level is 2 to indicate a mid-range configuration car.
  • the convolutional neural network to be trained is trained through a large amount of data, a convolutional neural network model for identifying the vehicle configuration level is obtained.
  • step S110 includes:
  • S111 Perform grayscale processing on the vehicle picture to be identified to obtain a grayscale image
  • S112 Perform edge detection on the grayscale image to obtain a grayscale image after edge detection
  • the picture of the vehicle to be identified is generally a color picture.
  • Color images contain more information, but if you process the color images directly, the execution speed will be reduced and the storage space will increase.
  • the grayscale of color pictures is a basic method of image processing. It is widely used in the field of pattern recognition. A reasonable grayscale will be extremely helpful for the extraction and subsequent processing of image information, saving storage space and speeding up processing. speed.
  • the method of edge detection is to examine the changes in the gray level of the pixels of the image in a certain area, and to identify the points with obvious brightness changes in the digital image.
  • Image edge detection can greatly reduce the amount of data, and eliminate irrelevant information, and preserve the important structural attributes of the image.
  • operators used for edge detection the commonly used ones are Sobel operator (ie Sobel operator), Laplacian edge detection operator (ie Laplacian edge detection operator), Canny edge detection operator (ie Hunk Edge detection operator) and so on.
  • the nameplate image Perform binarization which is a type of image thresholding. According to the selection of the threshold, the binarization method can be divided into global threshold method, dynamic threshold method and local threshold method.
  • the maximum between-class variance method also called Otsu algorithm
  • Otsu algorithm is commonly used for thresholding to eliminate some of the smaller gradient values Pixels, reduce the range of the license plate that needs to be searched, the pixel value of the license plate image after binarization is 0 or 255.
  • step S115 includes:
  • the candidate area of the frame number nameplate is located, and the candidate area of the frame number nameplate is divided by the license plate characters to obtain the frame number.
  • License plate character segmentation is to segment the license plate characters in the license plate area that has been located, so as to obtain the characters on the license plate, which is the premise and preparation of the license plate character recognition.
  • the vehicle model can be obtained according to the VIN code.
  • the first digit indicates the country code of production; the second digit indicates the manufacturer code; the third digit indicates the model and type code; the fourth to sixth digits indicate the model code; the seventh digit indicates the engine Model code; the eighth digit represents the occupant safety protection device code; the ninth digit represents the VIN inspection number code; the tenth digit represents the model year code; the eleventh digit represents the assembly factory code; the twelfth digit represents the factory sequence number code;
  • the vehicle model can be obtained from the fourth to sixth digits above.
  • dividing the candidate area of the frame number and nameplate by license plate characters in step S115 to obtain the frame number includes:
  • the relative height is determined by vertical projection.
  • the two adjacent connected domains are merged to obtain the candidate area of the frame number plate after the merge process;
  • the method of combining connected domains and projection is used to segment the frame number nameplate candidate area, that is, the four-connected marking method is used to segment the frame number nameplate candidate area. Mark the boundary of the candidate area of the number plate to form a connected domain; then determine whether the height and width of each connected domain is basically equal to the height and width of the candidate area of the frame number plate (which has been calculated when the frame is removed).
  • each connected domain corresponding to the candidate area of the frame number plate is added with a rectangular frame, and a single frame number character is extracted.
  • S120 Identify and acquire each vehicle interior component in the image of the vehicle to be identified to obtain a corresponding sequence of vehicle interior components.
  • step S120 includes:
  • S121 Identify and acquire the steering wheel, sunroof, transmission, seat, and air outlet of the air conditioner in the image of the vehicle to be identified;
  • S122 According to a preset labeling strategy, obtain label values corresponding to the steering wheel, sunroof, transmission, seat, and air outlet of the air conditioner in the image of the vehicle to be identified to form a sequence of vehicle interior parts.
  • the labeling strategy is 1 when it is a multi-function steering wheel, and 0 when it is not a multi-function steering wheel; when there is a sunroof, it is 1 and when there is no sunroof, it is 0; when it is an automatic transmission, it is 1 and it is manual
  • the transmission is marked as 0; when it is a leather seat, it is marked as 1, and when it is not a leather seat, it is marked as 0; when there is an air conditioning outlet in the rear, it is marked as 1, and when there is no air conditioning outlet in the rear, it is marked as 0; for example, After a vehicle is marked as [1,1,1,1], it means that the details of the vehicle’s interior parts are multi-function steering wheel, sunroof, automatic transmission, leather seats
  • the vehicle interior part sequence corresponding to the image of the vehicle to be identified is obtained, the vehicle interior part sequence is used as the input of the pre-trained convolutional neural network model to obtain the corresponding vehicle configuration level .
  • the corresponding vehicle configuration level (such as 2) can be obtained.
  • the convolutional neural network model effectively judges the vehicle configuration level, which is convenient for determining the actual value of the vehicle in the process of auto insurance claims.
  • S140 According to the vehicle model and vehicle configuration level corresponding to the vehicle image to be identified, obtain unique vehicle model information corresponding to the vehicle image to be identified, and store the unique vehicle model information in a preset storage area.
  • the unique model information corresponding to the vehicle image to be identified can be automatically acquired, and the vehicle can be estimated based on the vehicle model and vehicle configuration level.
  • the actual value is obtained through image recognition, without the user's manual input.
  • This method realizes the efficient automatic identification and entry of vehicle models and avoids the risk of data errors caused by manual entry of vehicle information.
  • the embodiment of the present application also provides a vehicle type identification device, which is used to execute any embodiment of the aforementioned vehicle type identification method.
  • a vehicle type identification device which is used to execute any embodiment of the aforementioned vehicle type identification method.
  • FIG. 6, is a schematic block diagram of a vehicle type identification device provided by an embodiment of the present application.
  • the vehicle type recognition device 100 may be configured in a server.
  • the vehicle model recognition device 100 includes a vehicle model acquisition unit 110, an input sequence acquisition unit 120, a configuration level acquisition unit 130, and a vehicle model recognition unit 140.
  • the vehicle model acquisition unit 110 is configured to acquire the vehicle model according to the frame number in the vehicle image to be identified if the image of the vehicle to be identified is detected and the image of the vehicle to be identified includes vehicle interior parts.
  • the uploader such as smart phones, tablet computers, and other smart terminals
  • the uploader takes pictures of the vehicle from multiple perspectives (especially pictures of the interior of the vehicle)
  • the server receives and detects the image of the vehicle to be identified, and the image of the vehicle to be identified includes vehicle interior parts, obtain the vehicle model according to the frame number in the image of the vehicle to be identified. Through the identification of the frame number to determine the vehicle model, the value estimation information of the vehicle can be quickly obtained. If the vehicle needs to be accurately estimated, the vehicle interior components need to be further identified.
  • the frame number also known as Vehicle Identification Number (Vehicle Identification Number, abbreviated as VIN)
  • VIN Vehicle Identification Number
  • the frame number is a set of seventeen alphanumeric numbers, a set of unique numbers used on cars, and can identify the manufacturer of the car , Engine, chassis serial number and other performance information. To avoid confusion with the number 1,0, the English letters "I”, "O", and "Q” will not be used.
  • the VIN code of a car can mostly be found on the left side of the dashboard and under the windshield, usually written on a nameplate. Therefore, the process of recognizing the frame number in the picture of the vehicle to be recognized is similar to the process of license plate recognition.
  • the vehicle type recognition device 100 further includes:
  • the picture crawling unit 101 is configured to crawl through a crawler tool to crawl a web page corresponding to a preset URL list including vehicle pictures of vehicle interiors;
  • the model training unit 102 is configured to label each vehicle interior part in the vehicle picture as the input of the convolutional neural network to be trained, and use the vehicle configuration level corresponding to the vehicle picture as the output of the convolutional neural network to be trained,
  • the convolutional neural network to be trained is trained to obtain a convolutional neural network model for identifying the vehicle configuration level.
  • the vehicle model acquisition unit 110 includes:
  • the gray-scale unit 111 is used to perform gray-scale processing on the image of the vehicle to be identified to obtain a gray-scale image
  • the edge detection unit 112 is configured to perform edge detection on the gray image to obtain a gray image after edge detection;
  • the binarization unit 113 is configured to perform binarization processing on the gray image after edge detection to obtain a binarized gray image
  • the filtering unit 114 is configured to filter the binarized grayscale image to obtain the candidate area of the frame number plate;
  • the area locating unit 115 is used to locate the area including characters in the candidate area of the frame number nameplate to obtain the frame number;
  • the frame number analysis unit 116 is used to analyze the frame number to obtain the vehicle model.
  • the area positioning unit 115 is further configured to:
  • the candidate area of the frame number nameplate is located, and the candidate area of the frame number nameplate is divided by the license plate characters to obtain the frame number.
  • the input sequence acquiring unit 120 is used to identify and acquire each vehicle interior component in the image of the vehicle to be identified to obtain a corresponding vehicle interior component sequence.
  • the input sequence obtaining unit 120 includes:
  • the picture designated part recognition unit 121 is used to recognize and acquire the steering wheel, sunroof, transmission, seat, and air outlet of the air conditioner in the picture of the vehicle to be recognized;
  • the label value acquisition unit 122 is configured to acquire the label values corresponding to the steering wheel, sunroof, transmission, seat, and air outlet of the air conditioner in the image of the vehicle to be identified according to a preset labeling strategy to form a vehicle interior component sequence.
  • the configuration level acquiring unit 130 is configured to use the vehicle interior part sequence corresponding to the vehicle picture to be recognized as the input of the pre-trained convolutional neural network model to obtain the vehicle configuration level corresponding to the vehicle picture to be recognized.
  • the vehicle type identification unit 140 is configured to obtain unique vehicle type information corresponding to the vehicle image to be identified according to the vehicle model and vehicle configuration level corresponding to the vehicle image to be identified, and store the unique vehicle type information in a preset storage area .
  • the unique model information corresponding to the vehicle image to be identified can be automatically acquired, and the vehicle can be estimated based on the vehicle model and vehicle configuration level. The actual value. And these information are all obtained through image recognition, without the user's manual input.
  • the device realizes efficient automatic identification and entry of vehicle models, avoiding the risk of data errors caused by manual entry of vehicle information.
  • the vehicle identification device described above can be implemented in the form of a computer program, and the computer program can be run on a computer device as shown in FIG. 10.
  • FIG. 10 is a schematic block diagram of a computer device according to an embodiment of the present application.
  • the computer device 500 is a server, and the server may be an independent server or a server cluster composed of multiple servers.
  • the computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.
  • the non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032.
  • the processor 502 can execute the vehicle type recognition method.
  • the processor 502 is used to provide computing and control capabilities, and support the operation of the entire computer device 500.
  • the internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503.
  • the processor 502 can execute the vehicle type identification method.
  • the network interface 505 is used for network communication, such as providing data information transmission.
  • the structure shown in FIG. 10 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied.
  • the specific computer device 500 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.
  • the processor 502 is configured to run a computer program 5032 stored in a memory, so as to implement the vehicle type identification method in the embodiment of the present application.
  • the embodiment of the computer device shown in FIG. 10 does not constitute a limitation on the specific configuration of the computer device.
  • the computer device may include more or less components than shown in the figure. Or combine certain components, or different component arrangements.
  • the computer device may only include a memory and a processor. In such an embodiment, the structures and functions of the memory and the processor are consistent with the embodiment shown in FIG. 10, and will not be repeated here.
  • the processor 502 may be a central processing unit (Central Processing Unit, CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (Digital Signal Processors, DSPs), Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor.
  • a computer-readable storage medium may be a non-volatile computer-readable storage medium.
  • the computer-readable storage medium stores a computer program, where the computer program is executed by a processor to implement the vehicle type identification method in the embodiment of the present application.
  • the storage medium is a physical, non-transitory storage medium, such as a U disk, a mobile hard disk, a read-only memory (Read-Only Memory, ROM), a magnetic disk, or an optical disk that can store program codes. medium.
  • a physical, non-transitory storage medium such as a U disk, a mobile hard disk, a read-only memory (Read-Only Memory, ROM), a magnetic disk, or an optical disk that can store program codes. medium.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Evolutionary Biology (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

本申请公开了车型识别方法、装置、计算机设备及存储介质。该方法包括:若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号;识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列;将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级;根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息。

Description

车型识别方法、装置、计算机设备及存储介质
本申请要求于2019年3月15日提交中国专利局、申请号为201910197516.2、申请名称为“车型识别方法、装置、计算机设备及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及图像识别技术领域,尤其涉及一种车型识别方法、装置、计算机设备及存储介质。
背景技术
目前,汽车的种类和型号是日益增多,而且各种汽车的配置各不相同,这就增大了识别车辆的唯一车型的难度。例如在车险投保或理赔的过程中,在登记车辆的车型时,往往需要通过用户手动输入相关车型细节信息,这种方式耗时,而且容易由于人为的因素导致车型信息登记不准确。
申请内容
本申请实施例提供了一种车型识别方法、装置、计算机设备及存储介质,旨在解决现有技术中登记车辆的车型时是手动输入相关车型细节信息,不仅耗时,而且容易由于人为的因素导致车型信息登记不准确的问题。
第一方面,本申请实施例提供了一种车型识别方法,其包括:
若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号;
识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列;
将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级;
根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息。
第二方面,本申请实施例提供了一种车型识别装置,其包括:
车辆型号获取单元,用于若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号;
输入序列获取单元,用于识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列;
配置等级获取单元,用于将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级;
车型识别单元,用于根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息。
第三方面,本申请实施例又提供了一种计算机设备,其包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序,所述处理器执行所述计算机程序时实现上述第一方面所述的车型识别方法。
第四方面,本申请实施例还提供了一种计算机可读存储介质,其中所述计算机可读存储介质存储有计算机程序,所述计算机程序当被处理器执行时使所述处理器执行上述第一方面所述的车型识别方法。
附图说明
为了更清楚地说明本申请实施例技术方案,下面将对实施例描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的车型识别方法的应用场景示意图;
图2为本申请实施例提供的车型识别方法的流程示意图;
图3为本申请实施例提供的车型识别方法的另一流程示意图;
图4为本申请实施例提供的车型识别方法的子流程示意图;
图5为本申请实施例提供的车型识别方法的另一子流程示意图;
图6为本申请实施例提供的车型识别装置的示意性框图;
图7为本申请实施例提供的车型识别装置的另一示意性框图;
图8为本申请实施例提供的车型识别装置的子单元示意性框图;
图9为本申请实施例提供的车型识别装置的另一子单元示意性框图;
图10为本申请实施例提供的计算机设备的示意性框图。
具体实施方式
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
应当理解,当在本说明书和所附权利要求书中使用时,术语“包括”和“包含”指示所描述特征、整体、步骤、操作、元素和/或组件的存在,但并不排除一个或多个其它特征、整体、步骤、操作、元素、组件和/或其集合的存在或添加。
还应当理解,在此本申请说明书中所使用的术语仅仅是出于描述特定实施例的目的而并不意在限制本申请。如在本申请说明书和所附权利要求书中所使用的那样,除非上下文清楚地指明其它情况,否则单数形式的“一”、“一个”及“该”意在包括复数形式。
还应当进一步理解,在本申请说明书和所附权利要求书中使用的术语“和/或”是指相关联列出的项中的一个或多个的任何组合以及所有可能组合,并且包括这些组合。
请参阅图1和图2,图1为本申请实施例提供的车型识别方法的应用场景示意图,图2为本申请实施例提供的车型识别方法的流程示意图,该车型识别方法应用于服务器中,该方法通过安装于服务器中的应用软件进行执行。
如图2所示,该方法包括步骤S110~S140。
S110、若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号。
在本实施例中,是上传端(如智能手机、平板电脑等智能终端)在拍摄了多个视角的车辆图片(尤其是车辆内部的图片)时,将所拍摄并选定上传的车辆图片作为待识别车辆图片上传至服务器。若服务器接收并检测到待识别车辆图片,且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号。通过对车架号的识别以判断车辆型号,能快速的获知车辆的价值预估信息,如需对车辆进行准确估值,还需进一步识别车辆内饰部件。
其中,车架号,又称车辆识别号码(Vehicle Identification Number,简记为VIN),是一组由十七个英数组成,用于汽车上的一组唯一的号码,可以识别汽车的生产商、引擎、底盘序号及其他性能等资料。为避免与数字的1,0混淆,英文字母“I”、“O”、“Q”均不会被使用。轿车的VIN码大多可以在仪表板左侧、风挡玻璃下面找到,一般写在一块铭牌上。故对待识别车辆图片中的车架号进行识别,其过程类似于车牌识别过程。
在一实施例中,如图3所示,步骤S110之前还包括:
S101、通过爬虫工具爬取预设的网址列表对应网页中包括车辆内饰的车辆图片;
S102、将所述车辆图片中的各车辆内饰部件进行标注以作为待训练卷积神经网络的输入,将车辆图片对应的车辆配置等级作为待训练卷积神经网络的输出,对所述待训练卷积神经网络进行训练,得到用于识别车辆配置等级的卷积神经网络模型。
在本实施例中,通过爬虫工具从预设的网址列表(这些网址列表中存储了大量包括车辆内饰细节的车辆图片),此时可以将所爬取的车辆图片中所包括的内饰部件进行标注。
对车辆内饰部件进行标注时,主要关注:判断方向盘是否多功能,判断天窗有无,判断变速器是手动还是自动,判断是否真皮座椅(判断是否真皮座椅可以通过拍到的座椅图片中座椅部分的反光度来判断,真皮一般反光感不强,而非真皮座椅的拍出来的照片中心部分反光现象明显,会出现白花花的一片),判断后排是否有空调出风口等。例如,当是多功能方向盘时标注为1,不是多功能方向盘时标注为0;当有天窗时标注为1,无天窗时标注为0;当为自动变速器时标注为1,为手动变速器时标注为0;当为真皮座椅标注为1,不为真皮座椅时标注为0;当后排有空调出风口时标注为1,后排无空调出风口时标注为0;例如,某一车辆进行标注后为[1,1,1,1,1],则表示该车辆内饰部件细节为是多功能方向盘、有天窗、是自动变速器、是真皮座椅、后排有空调出风口;当对所爬取的车辆图片进行上述标注后,还需对车辆配置等级进行标注例如[1]、[2]、[3],其中1表示车辆配置等级为高档配置、2表示车辆配置等级为中档配置、3表示车辆配置等级为低档配置。例如某一车辆进行标注后为[1,1,1,1,1],其对应的车辆配置等级为1以表示高档配置车;例如某一车辆进行标注后为[0,0,0,0,0], 其对应的车辆配置等级为3以表示低档配置车;例如某一车辆进行标注后为[0,1,0,1,0],其对应的车辆配置等级为2以表示中档配置车。当通过大量的数据对所述待训练卷积神经网络进行训练,得到用于识别车辆配置等级的卷积神经网络模型。
在一实施例中,如图4所示,步骤S110包括:
S111、将所述待识别车辆图片进行灰度化处理,以得到灰度图;
S112、将所述灰度图进行边缘检测,得到边缘检测后灰度图;
S113、将所述边缘检测后灰度图进行二值化处理,得到二值化灰度图;
S114、对所述二值化灰度图进行滤波,得到车架号铭牌候选区域;
S115、定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号;
S116、解析所述车架号以获取车辆型号。
在本实施例中,将所述待识别车辆图片一般是彩色图片。彩色图片包含更多的信息,但是直接对彩色图像进行处理的话,执行速度将会降低,储存空间也会变大。彩色图片的灰度化是图像处理的一种基本的方法,在模式识别领域得到广泛的运用,合理的灰度化将对图像信息的提取和后续处理极有帮助,能够节省储存空间,加快处理速度。
边缘检测的方法是考察图像的像素在某个领域内灰度的变化情况,标识数字图像中亮度变化明显的点。图像的边缘检测能够大幅度地减少数据量,并且剔除不相关的信息,保存图像重要的结构属性。用于边缘检测的算子很多,常用的有Sobel算子(即索贝尔算子),还有Laplacian边缘检测算子(即拉普拉斯边缘检测算子)、Canny边缘检测算子(即坎尼边缘检测算子)等。
灰度图经过边缘检测之后,车架号铭牌上的字符及边缘信息会突出出来,同时,其他非字符和非铭牌边框的边缘纹理特征也突出了出来,为了减少噪声的影响,需要对铭牌图像进行二值化处理,二值化是对图像进行阈值化的一种类型。根据阈值的选取情况,二值化的方法可分为全局阈值法、动态阈值法和局部阈值法,常用最大类间方差法(也称Otsu算法)进行阈值化,来剔除一些梯度值较小的像素,减少需要查找的车牌范围,二值化处理后车牌图像的像素值为0或者255。
之后,在尽量保留图像细节特征的条件下对目标图像的噪声进行抑制,是图像处理中消除噪声的不可或缺的操作,其处理的结果的好坏将直接影响到对 后续图像进行处理和分析的有效性和可靠性。常用的滤波操作方法有很多种,如中值滤波、形态学滤波、高斯滤波、双边滤波等。通过对所述二值化灰度图进行滤波,得到车架号铭牌候选区域。
在一实施例中,步骤S115包括:
定位所述车架号铭牌候选区域,通过车牌字符分割所述车架号铭牌候选区域以获取车架号。
即在获取了车架号铭牌候选区域后,可参考车牌字符分割算法,以识别车架号铭牌候选区域的车架号。车牌字符分割就是对已经定位出的车牌区域内的车牌字符进行分割,从而获取车牌上的字符,是车牌字符识别的前提和准备。
当获取了车架号即可根据VIN码获取车辆型号。例如德国宝马汽车公司轿车VIN码中,第一位表示生产国别代码;第二位表示生产厂家代码;第三位表示车型及种类代码;第四~六位表示车型代码;第七位表示发动机型号代码;第八位表示乘员安全保护装置代码;第九位表示VIN检验数代码;第十位表示车型年款代码;第十一位表示总装工厂代码;第十二位表示出厂顺序号代码;通过上述第四~六位即可获取车辆型号。
在一实施例中,步骤S115中通过车牌字符分割所述车架号铭牌候选区域以获取车架号包括:
通过四连通标记对车架号铭牌候选区域进行标记,形成连通域;
判断各连通域的高度是否等于车架号铭牌候选区域的高度,且判断各连通域的宽度是否等于车架号铭牌候选区域的宽度;
若有相邻的两个连通域的高度均小于车架号铭牌候选区域的高度、且该相邻的两个连通域的宽度均小于车架号铭牌候选区域的宽度,通过垂直投影将该相邻的两个连通域进行合并,得到合并处理后车架号铭牌候选区域;
若有连通域的高度大于车架号铭牌候选区域的高度、且该连通域的宽度大于车架号铭牌候选区域的宽度,通过垂直投影将该连通域进行分割,得到分割处理后车架号铭牌候选区域;
通过对合并处理后车架号铭牌候选区域或分割处理后车架号铭牌候选区域的各个连通域添加矩形边框,得到车架号铭牌候选区域对应的多个车架号字符;
通过对多个车架号字符分别进行图像识别,得到对应的车架号。
在本实施例中,通过车牌字符分割所述车架号铭牌候选区域时,采用连通 域和投影相结合的方法来对车架号铭牌候选区域进行字符分割,即采用四连通标记法对车架号铭牌候选区域的边界进行标记,形成连通域;然后判断各个连通域的高宽是否基本等于车架号铭牌候选区域的高宽(去边框时已经求出),若相差较大时,就进行垂直投影,把宽小于车架号铭牌候选区域宽的相邻区域进行合并,把宽大于车架号铭牌候选区域宽的相邻区域进行进一步分割;最后对合并处理后车架号铭牌候选区域或分割处理后车架号铭牌候选区域对应的各个连通域加矩形边框,提取单个车架号字符。
S120、识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列。
在一实施例中,如图5所示,步骤S120包括:
S121、识别获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口;
S122、根据预设的标注策略,获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口分别对应的标示值以组成车辆内饰部件序列。
在本实施例中,当识别获取了待识别车辆图片中的各车辆内饰部件后,针对所列举的5个判断标准一一核对(这5个判断标准即为预先设置的标注策略)以得到对应的车辆内饰部件序列。其中,标注策略为当是多功能方向盘时标注为1,不是多功能方向盘时标注为0;当有天窗时标注为1,无天窗时标注为0;当为自动变速器时标注为1,为手动变速器时标注为0;当为真皮座椅标注为1,不为真皮座椅时标注为0;当后排有空调出风口时标注为1,后排无空调出风口时标注为0;例如,某一车辆进行标注后为[1,1,1,1,1],则表示该车辆内饰部件细节为是多功能方向盘、有天窗、是自动变速器、是真皮座椅、后排有空调出风口。
S130、将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级。
在本实施例中,当获取了与所述待识别车辆图片对应的车辆内饰部件序列,将车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,即可得到对应的车辆配置等级。例如,此时将[10100]输入至卷积神经网络模型,可得到对应的车辆配置等级(如2)。通过卷积神经网络模型有效的判断了车辆配置等级,便于车险理赔的过程中,确定车辆实际价值。
S140、根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息,将所述唯一车型信息存储至预设的存储区域。
在本实施例中,当获取了待识别车辆图片对应的车辆型号及车辆配置等级后,即可自动获取与所述待识别车辆图片对应的唯一车型信息,并根据车辆型号及车辆配置等级估算车辆的实际价值。而且这些信息均是通过图像识别来获取,无需用户手工录入。
该方法实现了车型的高效自动化识别和录入,避免了车型信息的人工录入而造成数据错误的风险。
本申请实施例还提供一种车型识别装置,该车型识别装置用于执行前述车型识别方法的任一实施例。具体地,请参阅图6,图6是本申请实施例提供的车型识别装置的示意性框图。该车型识别装置100可以配置于服务器中。
如图6所示,车型识别装置100包括车辆型号获取单元110、输入序列获取单元120、配置等级获取单元130、车型识别单元140。
车辆型号获取单元110,用于若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号。
在本实施例中,是上传端(如智能手机、平板电脑等智能终端)在拍摄了多个视角的车辆图片(尤其是车辆内部的图片)时,将所拍摄并选定上传的车辆图片作为待识别车辆图片上传至服务器。若服务器接收并检测到待识别车辆图片,且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号。通过对车架号的识别以判断车辆型号,能快速的获知车辆的价值预估信息,如需对车辆进行准确估值,还需进一步识别车辆内饰部件。
其中,车架号,又称车辆识别号码(Vehicle Identification Number,简记为VIN),是一组由十七个英数组成,用于汽车上的一组唯一的号码,可以识别汽车的生产商、引擎、底盘序号及其他性能等资料。为避免与数字的1,0混淆,英文字母“I”、“O”、“Q”均不会被使用。轿车的VIN码大多可以在仪表板左侧、风挡玻璃下面找到,一般写在一块铭牌上。故对待识别车辆图片中的车架号进行识别,其过程类似于车牌识别过程。
在一实施例中,如图7所示,车型识别装置100还包括:
图片爬取单元101,用于通过爬虫工具爬取预设的网址列表对应网页中包括车辆内饰的车辆图片;
模型训练单元102,用于将所述车辆图片中的各车辆内饰部件进行标注以作为待训练卷积神经网络的输入,将车辆图片对应的车辆配置等级作为待训练卷积神经网络的输出,对所述待训练卷积神经网络进行训练,得到用于识别车辆配置等级的卷积神经网络模型。
在一实施例中,如图8所示,车辆型号获取单元110包括:
灰度化单元111,用于将所述待识别车辆图片进行灰度化处理,以得到灰度图;
边缘检测单元112,用于将所述灰度图进行边缘检测,得到边缘检测后灰度图;
二值化单元113,用于将所述边缘检测后灰度图进行二值化处理,得到二值化灰度图;
滤波单元114,用于对所述二值化灰度图进行滤波,得到车架号铭牌候选区域;
区域定位单元115,用于定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号;
车架号解析单元116,用于解析所述车架号以获取车辆型号。
在一实施例中,区域定位单元115还用于:
定位所述车架号铭牌候选区域,通过车牌字符分割所述车架号铭牌候选区域以获取车架号。
输入序列获取单元120,用于识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列。
在一实施例中,如图9所示,输入序列获取单元120包括:
图片指定部位识别单元121,用于识别获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口;
标示值获取单元122,用于根据预设的标注策略,获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口分别对应的标示值以组成车辆内饰部件序列。
配置等级获取单元130,用于将与所述待识别车辆图片对应的车辆内饰部件 序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级。
车型识别单元140,用于根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息,将所述唯一车型信息存储至预设的存储区域。
在本实施例中,当获取了待识别车辆图片对应的车辆型号及车辆配置等级后,即可自动获取与所述待识别车辆图片对应的唯一车型信息,并根据车辆型号及车辆配置等级估算车辆的实际价值。而且这些信息均是通过图像识别来获取,无需用户手工录入。
该装置实现了车型的高效自动化识别和录入,避免了车型信息的人工录入而造成数据错误的风险。
上述车型识别装置可以实现为计算机程序的形式,该计算机程序可以在如图10所示的计算机设备上运行。
请参阅图10,图10是本申请实施例提供的计算机设备的示意性框图。该计算机设备500是服务器,服务器可以是独立的服务器,也可以是多个服务器组成的服务器集群。
参阅图10,该计算机设备500包括通过系统总线501连接的处理器502、存储器和网络接口505,其中,存储器可以包括非易失性存储介质503和内存储器504。
该非易失性存储介质503可存储操作系统5031和计算机程序5032。该计算机程序5032被执行时,可使得处理器502执行车型识别方法。
该处理器502用于提供计算和控制能力,支撑整个计算机设备500的运行。
该内存储器504为非易失性存储介质503中的计算机程序5032的运行提供环境,该计算机程序5032被处理器502执行时,可使得处理器502执行车型识别方法。
该网络接口505用于进行网络通信,如提供数据信息的传输等。本领域技术人员可以理解,图10中示出的结构,仅仅是与本申请方案相关的部分结构的框图,并不构成对本申请方案所应用于其上的计算机设备500的限定,具体的计算机设备500可以包括比图中所示更多或更少的部件,或者组合某些部件,或者具有不同的部件布置。
其中,所述处理器502用于运行存储在存储器中的计算机程序5032,以实现本申请实施例中车型识别方法。
本领域技术人员可以理解,图10中示出的计算机设备的实施例并不构成对计算机设备具体构成的限定,在其他实施例中,计算机设备可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件布置。例如,在一些实施例中,计算机设备可以仅包括存储器及处理器,在这样的实施例中,存储器及处理器的结构及功能与图10所示实施例一致,在此不再赘述。
应当理解,在本申请实施例中,处理器502可以是中央处理单元(Central Processing Unit,CPU),该处理器502还可以是其他通用处理器、数字信号处理器(Digital Signal Processor,DSP)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现成可编程门阵列(Field-Programmable Gate Array,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。其中,通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
在本申请的另一实施例中提供计算机可读存储介质。该计算机可读存储介质可以为非易失性的计算机可读存储介质。该计算机可读存储介质存储有计算机程序,其中计算机程序被处理器执行时实现本申请实施例中车型识别方法。
所述存储介质为实体的、非瞬时性的存储介质,例如可以是U盘、移动硬盘、只读存储器(Read-Only Memory,ROM)、磁碟或者光盘等各种可以存储程序代码的实体存储介质。
所属领域的技术人员可以清楚地了解到,为了描述的方便和简洁,上述描述的设备、装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
以上所述,仅为本申请的具体实施方式,但本申请的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本申请揭露的技术范围内,可轻易想到各种等效的修改或替换,这些修改或替换都应涵盖在本申请的保护范围之内。因此,本申请的保护范围应以权利要求的保护范围为准。

Claims (20)

  1. 一种车型识别方法,包括:
    若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号;
    识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列;
    将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级;
    根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息,将所述唯一车型信息存储至预设的存储区域。
  2. 根据权利要求1所述的车型识别方法,其中,所述若检测到待识别车辆图片且待识别车辆图片包括车辆内饰部件,根据待识别车辆图片中的车架号获取车辆型号之前,还包括:
    通过爬虫工具爬取预设的网址列表对应网页中包括车辆内饰的车辆图片;
    将所述车辆图片中的各车辆内饰部件进行标注以作为待训练卷积神经网络的输入,将车辆图片对应的车辆配置等级作为待训练卷积神经网络的输出,对所述待训练卷积神经网络进行训练,得到用于识别车辆配置等级的卷积神经网络模型。
  3. 根据权利要求1所述的车型识别方法,其中,所述根据所述待识别车辆图片中的车架号获取车辆型号,包括:
    将所述待识别车辆图片进行灰度化处理,以得到灰度图;
    将所述灰度图进行边缘检测,得到边缘检测后灰度图;
    将所述边缘检测后灰度图进行二值化处理,得到二值化灰度图;
    对所述二值化灰度图进行滤波,得到车架号铭牌候选区域;
    定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号;
    解析所述车架号以获取车辆型号。
  4. 根据权利要求3所述的车型识别方法,其中,所述定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号,包括:
    定位所述车架号铭牌候选区域,通过车牌字符分割所述车架号铭牌候选区域以获取车架号。
  5. 根据权利要求1所述的车型识别方法,其中,所述识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列,包括:
    识别获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口;
    根据预设的标注策略,获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口分别对应的标示值以组成车辆内饰部件序列。
  6. 根据权利要求5所述的车型识别方法,其中,所述根据预设的标注策略,获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口分别对应的标示值,包括:
    当是多功能方向盘时标注为1,不是多功能方向盘时标注为0;当有天窗时标注为1,无天窗时标注为0;当为自动变速器时标注为1,为手动变速器时标注为0;当为真皮座椅标注为1,不为真皮座椅时标注为0;当后排有空调出风口时标注为1,后排无空调出风口时标注为0。
  7. 根据权利要求4所述的车型识别方法,其中,所述通过车牌字符分割所述车架号铭牌候选区域以获取车架号,包括:
    通过四连通标记对车架号铭牌候选区域进行标记,形成连通域;
    判断各连通域的高度是否等于车架号铭牌候选区域的高度,且判断各连通域的宽度是否等于车架号铭牌候选区域的宽度;
    若有相邻的两个连通域的高度均小于车架号铭牌候选区域的高度、且该相邻的两个连通域的宽度均小于车架号铭牌候选区域的宽度,通过垂直投影将该相邻的两个连通域进行合并,得到合并处理后车架号铭牌候选区域;
    若有连通域的高度大于车架号铭牌候选区域的高度、且该连通域的宽度大于车架号铭牌候选区域的宽度,通过垂直投影将该连通域进行分割,得到分割处理后车架号铭牌候选区域;
    通过对合并处理后车架号铭牌候选区域或分割处理后车架号铭牌候选区域的各个连通域添加矩形边框,得到车架号铭牌候选区域对应的多个车架号字符;
    通过对多个车架号字符分别进行图像识别,得到对应的车架号。
  8. 一种车型识别装置,包括:
    车辆型号获取单元,用于若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号;
    输入序列获取单元,用于识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列;
    配置等级获取单元,用于将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级;
    车型识别单元,用于根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息,将所述唯一车型信息存储至预设的存储区域。
  9. 根据权利要求8所述的车型识别装置,其中,还包括:
    图片爬取单元,用于通过爬虫工具爬取预设的网址列表对应网页中包括车辆内饰的车辆图片;
    模型训练单元,用于将所述车辆图片中的各车辆内饰部件进行标注以作为待训练卷积神经网络的输入,将车辆图片对应的车辆配置等级作为待训练卷积神经网络的输出,对所述待训练卷积神经网络进行训练,得到用于识别车辆配置等级的卷积神经网络模型。
  10. 根据权利要求8所述的车型识别装置,其中,所述车辆型号获取单元,包括:
    灰度化单元,用于将所述待识别车辆图片进行灰度化处理,以得到灰度图;
    边缘检测单元,用于将所述灰度图进行边缘检测,得到边缘检测后灰度图;
    二值化单元,用于将所述边缘检测后灰度图进行二值化处理,得到二值化灰度图;
    滤波单元,用于对所述二值化灰度图进行滤波,得到车架号铭牌候选区域;
    区域定位单元,用于定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号;
    车架号解析单元,用于解析所述车架号以获取车辆型号。
  11. 一种计算机设备,包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序,所述处理器执行所述计算机程序时实现以下步骤:
    若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号;
    识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列;
    将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级;
    根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息,将所述唯一车型信息存储至预设的存储区域。
  12. 根据权利要求11所述的计算机设备,其中,所述若检测到待识别车辆图片且待识别车辆图片包括车辆内饰部件,根据待识别车辆图片中的车架号获取车辆型号之前,还包括:
    通过爬虫工具爬取预设的网址列表对应网页中包括车辆内饰的车辆图片;
    将所述车辆图片中的各车辆内饰部件进行标注以作为待训练卷积神经网络的输入,将车辆图片对应的车辆配置等级作为待训练卷积神经网络的输出,对所述待训练卷积神经网络进行训练,得到用于识别车辆配置等级的卷积神经网络模型。
  13. 根据权利要求11所述的计算机设备,其中,所述根据所述待识别车辆图片中的车架号获取车辆型号,包括:
    将所述待识别车辆图片进行灰度化处理,以得到灰度图;
    将所述灰度图进行边缘检测,得到边缘检测后灰度图;
    将所述边缘检测后灰度图进行二值化处理,得到二值化灰度图;
    对所述二值化灰度图进行滤波,得到车架号铭牌候选区域;
    定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号;
    解析所述车架号以获取车辆型号。
  14. 根据权利要求13所述的计算机设备,其中,所述定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号,包括:
    定位所述车架号铭牌候选区域,通过车牌字符分割所述车架号铭牌候选区域以获取车架号。
  15. 根据权利要求11所述的计算机设备,其中,所述识别获取所述待识别 车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列,包括:
    识别获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口;
    根据预设的标注策略,获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口分别对应的标示值以组成车辆内饰部件序列。
  16. 根据权利要求15所述的计算机设备,其中,所述根据预设的标注策略,获取所述待识别车辆图片中的方向盘、天窗、变速器、座椅、及空调出风口分别对应的标示值,包括:
    当是多功能方向盘时标注为1,不是多功能方向盘时标注为0;当有天窗时标注为1,无天窗时标注为0;当为自动变速器时标注为1,为手动变速器时标注为0;当为真皮座椅标注为1,不为真皮座椅时标注为0;当后排有空调出风口时标注为1,后排无空调出风口时标注为0。
  17. 根据权利要求14所述的计算机设备,其中,所述通过车牌字符分割所述车架号铭牌候选区域以获取车架号,包括:
    通过四连通标记对车架号铭牌候选区域进行标记,形成连通域;
    判断各连通域的高度是否等于车架号铭牌候选区域的高度,且判断各连通域的宽度是否等于车架号铭牌候选区域的宽度;
    若有相邻的两个连通域的高度均小于车架号铭牌候选区域的高度、且该相邻的两个连通域的宽度均小于车架号铭牌候选区域的宽度,通过垂直投影将该相邻的两个连通域进行合并,得到合并处理后车架号铭牌候选区域;
    若有连通域的高度大于车架号铭牌候选区域的高度、且该连通域的宽度大于车架号铭牌候选区域的宽度,通过垂直投影将该连通域进行分割,得到分割处理后车架号铭牌候选区域;
    通过对合并处理后车架号铭牌候选区域或分割处理后车架号铭牌候选区域的各个连通域添加矩形边框,得到车架号铭牌候选区域对应的多个车架号字符;
    通过对多个车架号字符分别进行图像识别,得到对应的车架号。
  18. 一种计算机可读存储介质,所述计算机可读存储介质存储有计算机程序,所述计算机程序当被处理器执行时使所述处理器执行以下操作:
    若检测到待识别车辆图片且所述待识别车辆图片包括车辆内饰部件,根据所述待识别车辆图片中的车架号获取车辆型号;
    识别获取所述待识别车辆图片中的各车辆内饰部件,以得到对应的车辆内饰部件序列;
    将与所述待识别车辆图片对应的车辆内饰部件序列作为预先训练的卷积神经网络模型的输入,得到与所述待识别车辆图片对应的车辆配置等级;
    根据所述待识别车辆图片对应的车辆型号及车辆配置等级,获取与所述待识别车辆图片对应的唯一车型信息,将所述唯一车型信息存储至预设的存储区域。
  19. 根据权利要求18所述的计算机可读存储介质,其中,所述若检测到待识别车辆图片且待识别车辆图片包括车辆内饰部件,根据待识别车辆图片中的车架号获取车辆型号之前,还包括:
    通过爬虫工具爬取预设的网址列表对应网页中包括车辆内饰的车辆图片;
    将所述车辆图片中的各车辆内饰部件进行标注以作为待训练卷积神经网络的输入,将车辆图片对应的车辆配置等级作为待训练卷积神经网络的输出,对所述待训练卷积神经网络进行训练,得到用于识别车辆配置等级的卷积神经网络模型。
  20. 根据权利要求18所述的计算机可读存储介质,其中,所述根据所述待识别车辆图片中的车架号获取车辆型号,包括:
    将所述待识别车辆图片进行灰度化处理,以得到灰度图;
    将所述灰度图进行边缘检测,得到边缘检测后灰度图;
    将所述边缘检测后灰度图进行二值化处理,得到二值化灰度图;
    对所述二值化灰度图进行滤波,得到车架号铭牌候选区域;
    定位所述车架号铭牌候选区域中包括字符的区域,以获取车架号;
    解析所述车架号以获取车辆型号。
PCT/CN2019/118435 2019-03-15 2019-11-14 车型识别方法、装置、计算机设备及存储介质 WO2020186790A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910197516.2A CN110070092A (zh) 2019-03-15 2019-03-15 车型识别方法、装置、计算机设备及存储介质
CN201910197516.2 2019-03-15

Publications (1)

Publication Number Publication Date
WO2020186790A1 true WO2020186790A1 (zh) 2020-09-24

Family

ID=67366172

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/118435 WO2020186790A1 (zh) 2019-03-15 2019-11-14 车型识别方法、装置、计算机设备及存储介质

Country Status (2)

Country Link
CN (1) CN110070092A (zh)
WO (1) WO2020186790A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112734778A (zh) * 2021-01-28 2021-04-30 携程计算机技术(上海)有限公司 基于神经网络的车辆抠图方法、系统、设备及存储介质
CN114357769A (zh) * 2022-01-04 2022-04-15 一汽解放汽车有限公司 内饰覆盖件选材方法、装置、计算机设备、存储介质
CN115410209A (zh) * 2022-10-31 2022-11-29 山东济矿鲁能煤电股份有限公司阳城煤矿 一种基于图像处理的煤矿工单识别方法

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110070092A (zh) * 2019-03-15 2019-07-30 平安科技(深圳)有限公司 车型识别方法、装置、计算机设备及存储介质
CN112987676A (zh) * 2019-12-02 2021-06-18 北京金奔腾汽车科技有限公司 一种汽车车型智能识别系统及方法
CN111310561A (zh) * 2020-01-07 2020-06-19 成都睿琪科技有限责任公司 车辆配置识别方法及装置
CN113312945A (zh) * 2020-02-27 2021-08-27 浙江大搜车软件技术有限公司 车辆识别方法、装置、电子设备及可读存储介质
CN111652274B (zh) * 2020-04-28 2024-05-14 中国平安财产保险股份有限公司 基于图像识别的车型识别方法、装置及计算机设备
CN113744202A (zh) * 2021-08-13 2021-12-03 上海钧正网络科技有限公司 一种网约车的车辆一致性检测方法和装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105825218A (zh) * 2016-04-01 2016-08-03 深圳市元征科技股份有限公司 汽车车辆识别码的识别方法及装置
CN108122001A (zh) * 2017-12-13 2018-06-05 北京小米移动软件有限公司 图像识别方法及装置
CN108319907A (zh) * 2018-01-26 2018-07-24 腾讯科技(深圳)有限公司 一种车辆识别方法、装置和存储介质
CN109325491A (zh) * 2018-08-16 2019-02-12 腾讯科技(深圳)有限公司 识别码识别方法、装置、计算机设备和存储介质
CN110070092A (zh) * 2019-03-15 2019-07-30 平安科技(深圳)有限公司 车型识别方法、装置、计算机设备及存储介质

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102169633A (zh) * 2011-03-30 2011-08-31 成都兴同达微电子科技有限公司 机动车身份识别的方法
CN108573198A (zh) * 2017-03-14 2018-09-25 优信互联(北京)信息技术有限公司 一种根据车架号识别车型信息的方法及装置
CN108038423B (zh) * 2017-11-22 2022-03-04 广东数相智能科技有限公司 基于图像识别的汽车车型的识别方法及装置
CN108596277B (zh) * 2018-05-10 2020-07-07 腾讯科技(深圳)有限公司 一种车辆身份识别方法、装置和存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105825218A (zh) * 2016-04-01 2016-08-03 深圳市元征科技股份有限公司 汽车车辆识别码的识别方法及装置
CN108122001A (zh) * 2017-12-13 2018-06-05 北京小米移动软件有限公司 图像识别方法及装置
CN108319907A (zh) * 2018-01-26 2018-07-24 腾讯科技(深圳)有限公司 一种车辆识别方法、装置和存储介质
CN109325491A (zh) * 2018-08-16 2019-02-12 腾讯科技(深圳)有限公司 识别码识别方法、装置、计算机设备和存储介质
CN110070092A (zh) * 2019-03-15 2019-07-30 平安科技(深圳)有限公司 车型识别方法、装置、计算机设备及存储介质

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112734778A (zh) * 2021-01-28 2021-04-30 携程计算机技术(上海)有限公司 基于神经网络的车辆抠图方法、系统、设备及存储介质
CN114357769A (zh) * 2022-01-04 2022-04-15 一汽解放汽车有限公司 内饰覆盖件选材方法、装置、计算机设备、存储介质
CN114357769B (zh) * 2022-01-04 2024-06-04 一汽解放汽车有限公司 内饰覆盖件选材方法、装置、计算机设备、存储介质
CN115410209A (zh) * 2022-10-31 2022-11-29 山东济矿鲁能煤电股份有限公司阳城煤矿 一种基于图像处理的煤矿工单识别方法

Also Published As

Publication number Publication date
CN110070092A (zh) 2019-07-30

Similar Documents

Publication Publication Date Title
WO2020186790A1 (zh) 车型识别方法、装置、计算机设备及存储介质
CN111402264B (zh) 图像区域分割方法、装置及其模型训练方法及计算机设备
CN108682015B (zh) 一种生物图像中的病灶分割方法、装置、设备及存储介质
CN108629292B (zh) 弯曲车道线检测方法、装置及终端
EP3171295B1 (en) Method for processing image and computer-readable non-transitory recording medium storing program
CN112580734B (zh) 目标检测模型训练方法、系统、终端设备及存储介质
CN111507324B (zh) 卡片边框识别方法、装置、设备和计算机存储介质
CN104751097A (zh) 一种车辆识别码的检测处理方法及装置
CN111652208A (zh) 用户界面组件的识别方法、装置、电子设备及存储介质
CN114820679B (zh) 图像标注方法、装置、电子设备和存储介质
CN113537037A (zh) 路面病害识别方法、系统、电子设备及存储介质
CN112419207A (zh) 一种图像矫正方法及装置、系统
CN117094975A (zh) 钢铁表面缺陷检测方法、装置及电子设备
CN100561504C (zh) 文档的自适应二值化方法和设备
CN113902740A (zh) 图像模糊程度评价模型的构建方法
CN111898540A (zh) 车道线检测方法、装置、计算机设备及计算机可读存储介质
CN110135288B (zh) 一种电子证照的快速核对方法及装置
CN108090425B (zh) 一种车道线检测方法、装置及终端
CN111444911A (zh) 车牌识别模型的训练方法及装置、车牌识别方法及装置
US9858493B2 (en) Method and apparatus for performing registration plate detection with aid of edge-based sliding concentric windows
CN112837384A (zh) 车辆标记方法、装置和电子设备
CN110555344B (zh) 车道线识别方法、装置、电子设备、存储介质
CN112633289B (zh) 一种粘连字符分割方法和系统
CN111881897B (zh) 一种停车场地面中文标志识别方法及其系统、存储介质
CN114529570A (zh) 图像分割方法、图像识别方法、用户凭证补办方法及系统

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: 19919600

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19919600

Country of ref document: EP

Kind code of ref document: A1