Vehicle identification method, device, storage medium and system
By combining the data of photography equipment and lidar equipment, image processing and three-dimensional point cloud analysis technology are used to identify tiny objects on the carriage, solving the problem of difficulty in identification in the existing technology, and achieving more accurate vehicle type identification and Fault detection.
Patent Information
- Application Number
- CN202411177824.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-11-26
AI Technical Summary
The prior art is difficult to accurately identify the tiny identification objects (≤5cm) on the car during the vehicle self-inspection process. Due to the small target size, the lack of effective guidance and the performance limitations of image segmentation methods, the recognition difficulty increases.
Combining the carriage images collected by photography equipment and the three-dimensional point cloud data collected by lidar equipment, image processing algorithms, three-dimensional point cloud data analysis algorithms and neural network model technology are used to obtain the specific attributes, feature sets and self-test templates of the carriage to identify the carriage type.
It improves the accuracy of vehicle type identification, can position small objects in a moving state, enhances the pertinence of image detection, solves the problem of micro-identification objects recognition, and provides alternative judgment basis when the lighting conditions are insufficient.
Smart Images

Figure CN119027891A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of image recognition technology, and in particular to a vehicle recognition method, device, storage medium and system. Background Art
[0002] In the process of vehicle self-inspection, it is often difficult to accurately identify or locate some tiny objects (≤5cm) on large targets such as vehicles / carriages through image detection algorithms alone. On the one hand, because the target volume is too small and accounts for too small a proportion of the overall image, most detection algorithms will ignore it by default. On the other hand, it is also because traditional image segmentation methods lack effective target guidance, making it difficult to accurately locate small and effective targets. Finally, due to the performance limitations of image acquisition equipment, it is difficult to capture more details of small targets in motion while capturing image data over a large range, which further increases the difficulty of target recognition.
[0003] That is, the existing solutions cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a carriage. Summary of the invention
[0004] The main purpose of the present application is to provide a vehicle identification method, device, storage medium and system to at least solve the problem that the existing solutions cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a compartment.
[0005] To achieve the above-mentioned purpose, according to one aspect of the present application, a vehicle identification method is provided, wherein the vehicle has a carriage, and the method comprises: when the carriage of the vehicle enters the identification area, acquiring in real time a carriage image captured by a photographic device, and acquiring three-dimensional point cloud data of the carriage captured by a lidar device; using an image processing algorithm to analyze the carriage image, determine whether the quality of the carriage image meets the requirements, and determine feature set data corresponding to the carriage image or the three-dimensional point cloud data of the carriage, wherein the image processing algorithm is a plane determination algorithm, and the feature set data comprises at least one of the following: specific attributes, feature sets and self-test templates of the carriage, wherein the specific attributes include the length, width and height of the carriage, and the feature set includes the three-dimensional shape of the front of the vehicle and the blocking of the carriage. The height and width of the board, the self-inspection template includes an inspection template for whether the tarpaulin height is overloaded, an inspection template for whether the reinforcement is broken, and an inspection template for whether the car baffle is abnormal; when it is determined that the quality of the car image does not meet the requirements, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the car to determine whether there are preset abnormal points in the three-dimensional point cloud data of the car, and the preset abnormal points are raised or recessed abnormal points; when it is determined that the preset abnormal points exist in the three-dimensional point cloud data of the car, a neural network model technology is used to process the preset abnormal points and feature set data to obtain multiple type confidences, and the type of the vehicle is determined to be a type corresponding to the target type confidence, and the target type confidence is the maximum value of all the type confidences.
[0006] Optionally, an image processing algorithm is used to analyze the cabin image to determine whether the quality of the cabin image meets the requirements, including: using an image quality evaluation algorithm to score the cabin images at multiple moments to obtain multiple image scores; analyzing a target cabin image to determine whether the quality of the target cabin image meets the requirements, the target cabin image being the maximum value of all the image scores.
[0007] Optionally, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the car to determine whether there are preset abnormal points in the three-dimensional point cloud data of the car, including: using the three-dimensional point cloud data analysis algorithm to analyze the three-dimensional point cloud data of the target car to determine whether there are the preset abnormal points in the three-dimensional point cloud data of the car, the three-dimensional point cloud data of the target car is the three-dimensional point cloud data of the car at the same time of acquisition as the target car image.
[0008] Optionally, an image quality evaluation algorithm is used to score the vehicle compartment images at multiple moments to obtain multiple image scores, including: using the image quality evaluation algorithm to score the vehicle compartment images at multiple moments based on image clarity, the number of obstructions, color brightness and exposure to obtain multiple image scores.
[0009] Optionally, in the process of using neural network model technology to process the preset abnormal point and feature set data to obtain multiple types of confidence, the method also includes: when it is determined that the tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is abnormal, a first alarm message is generated to prompt that the tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is abnormal; when it is determined that the tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is normal, a second alarm message is generated to prompt that the tarpaulin is highly overloaded and the tie rods are broken; when it is determined that the tarpaulin is highly overloaded, the tie rods are not broken, and the car baffle is normal, a third alarm message is generated to prompt that the tarpaulin is highly overloaded.
[0010] Optionally, determining feature set data corresponding to the vehicle compartment image or the vehicle compartment three-dimensional point cloud data includes: when the resolution of the license plate number portion recognized on the vehicle compartment image is less than a preset resolution, determining the feature set data corresponding to the vehicle compartment three-dimensional point cloud data; when the resolution of the license plate number portion recognized on the vehicle compartment image is greater than or equal to a preset resolution, determining the feature set data corresponding to the vehicle compartment image.
[0011] Optionally, the method also includes: when it is determined that the specific attribute is within a preset length range, storing the coordinates of each point cloud of the car three-dimensional point cloud data; when it is determined that the specific attribute is not within the preset length range, determining not to store the coordinates of each point cloud of the car three-dimensional point cloud data.
[0012] According to another aspect of the present application, a vehicle identification device is provided, the vehicle having a carriage, the device comprising:
[0013] an acquisition unit, used to acquire, in real time, an image of the vehicle compartment acquired by a photographic device and three-dimensional point cloud data of the vehicle compartment acquired by a laser radar device when the vehicle compartment enters a recognition area;
[0014] A first determination unit is used to analyze the carriage image by using an image processing algorithm to determine whether the quality of the carriage image meets the requirements, and determine feature set data corresponding to the carriage image or the three-dimensional point cloud data of the carriage, wherein the image processing algorithm is a plane determination algorithm, and the feature set data includes at least one of the following: specific attributes of the carriage, a feature set, and a self-checking template, wherein the specific attributes include the length, width, and height of the carriage, the feature set includes the three-dimensional shape of the front of the vehicle and the height and width of the baffle of the carriage, and the self-checking template includes an inspection template for whether the tarpaulin height is overloaded, an inspection template for whether the tie rod is broken, and an inspection template for whether the carriage baffle is abnormal;
[0015] A second determination unit is used to analyze the three-dimensional point cloud data of the carriage using a three-dimensional point cloud data analysis algorithm when it is determined that the quality of the carriage image does not meet the requirements, and determine whether there is a preset abnormal point in the three-dimensional point cloud data of the carriage, wherein the preset abnormal point is a convex or concave abnormal point;
[0016] A processing unit is used to, when it is determined that the preset abnormal point exists in the three-dimensional point cloud data of the vehicle compartment, use a neural network model technology to process the preset abnormal point and feature set data to obtain multiple type confidences, and determine that the type of the vehicle is a type corresponding to a target type confidence, wherein the target type confidence is the maximum value of all the type confidences.
[0017] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the vehicle identification methods.
[0018] According to another aspect of the present application, a vehicle identification system is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include methods for executing any one of the vehicle identification methods.
[0019] By applying the technical solution of the present application, the images of the carriage are collected by photographic equipment and the three-dimensional point cloud data of the carriage are collected by lidar equipment, wherein a three-dimensional point cloud data analysis algorithm, a neural network model technology and an image processing algorithm are also adopted. At the same time, since the specific attributes, feature sets and self-inspection templates of the carriage are also taken into consideration, the targeted nature of image detection is enhanced, and the type of vehicle can be identified more accurately than the existing scheme, thereby solving the problem that the existing scheme cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a carriage. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0021] Figure 1 A schematic diagram of a process flow of a vehicle identification method provided according to an embodiment of the present application is shown;
[0022] Figure 2 A schematic flow chart of another vehicle identification method provided according to an embodiment of the present application is shown;
[0023] Figure 3 A structural block diagram of a vehicle identification device provided according to an embodiment of the present application is shown. DETAILED DESCRIPTION
[0024] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.
[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0027] For the convenience of description, some nouns or terms involved in the embodiments of the present application are explained below:
[0028] OpenCV contrib is an extension library containing various extended functions and algorithms. It provides some functions not available in the standard OpenCV library, such as feature detection and description, object recognition, image segmentation, etc. These functions and algorithms can help developers to implement various computer vision tasks more easily. The OpenCV contrib library exists as an additional module of OpenCV, and developers can selectively use it to expand their OpenCV projects.
[0029] The qualit module in OpenCV contrib is a module for image quality assessment. It contains a series of algorithms and tools for evaluating image quality, which can help users analyze and improve the quality of images. This module can be used to detect image blur, noise level, contrast and other indicators to help users evaluate the quality of images and perform necessary processing. These functions are very useful for image processing and computer vision applications.
[0030] As introduced in the background technology, in the process of vehicle self-inspection, some tiny identification objects (≤5cm) on large targets such as vehicles / carriages are often difficult to accurately identify or locate simply through image detection algorithms. On the one hand, because the target volume is indeed too small and accounts for too small a proportion of the overall image, most detection algorithms will ignore it by default. On the other hand, it is also because traditional image segmentation methods lack effective target guidance basis, which makes it difficult to accurately locate small effective targets. Finally, due to the performance limitations of image acquisition equipment, while capturing image data over a large range, it is difficult to capture more details for small targets in motion, thereby further increasing the difficulty of target recognition. In order to solve the problem that the existing solution cannot identify the type of vehicle based on the tiny identification objects (≤5cm) of a vehicle with a carriage, the embodiments of the present application provide a vehicle recognition method, device, storage medium and system.
[0031] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.
[0032] In this embodiment, a method for identifying a vehicle with a carriage is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0033] Figure 1 FIG. 1 is a flow chart of a vehicle identification method provided according to an embodiment of the present application. Figure 1 As shown, the method comprises the following steps:
[0034] Step S101, when the vehicle compartment enters the identification area, obtaining a vehicle compartment image captured by a photographic device in real time, and obtaining three-dimensional point cloud data of the vehicle compartment captured by a laser radar device;
[0035] The resolution of the camera device must be ≥1080p, and the number of lines of the above-mentioned lidar device must be ≥100 lines.
[0036] Step S102, using an image processing algorithm to analyze the above-mentioned carriage image, determine whether the quality of the above-mentioned carriage image meets the requirements, and determine the feature set data corresponding to the above-mentioned carriage image or the above-mentioned carriage three-dimensional point cloud data, the above-mentioned image processing algorithm is a plane determination algorithm, and the above-mentioned feature set data includes at least one of the following: specific attributes, feature sets and self-inspection templates of the above-mentioned carriage, the above-mentioned specific attributes include the length, width and height of the above-mentioned carriage, the above-mentioned feature set includes the three-dimensional shape of the front of the vehicle and the height and width of the baffle of the above-mentioned carriage, and the above-mentioned self-inspection template includes an inspection template for whether the tarpaulin height is overloaded, an inspection template for whether the tie rod is broken, and an inspection template for whether the carriage baffle is abnormal;
[0037] The step S102 uses an image processing algorithm to analyze the carriage image to determine whether the quality of the carriage image meets the requirements, including:
[0038] Using an image quality evaluation algorithm, the carriage images at multiple moments are scored to obtain multiple image scores;
[0039] Specifically, the above image quality evaluation algorithm (i.e., image quality evaluation is implemented based on the qualit module in OpenCV contrib) is used to score the above compartment images at multiple moments based on the image clarity, the number of occluders, the color brightness and the exposure, and obtain multiple image scores. By considering the image clarity, the number of occluders, the color brightness and the exposure, the score obtained by the image quality evaluation algorithm is more in line with the actual situation.
[0040] The target compartment image is analyzed to determine whether the quality of the target compartment image meets the requirements, and the target compartment image is the maximum value among all the image scores.
[0041] Determining the feature set data corresponding to the above-mentioned vehicle compartment image or the above-mentioned vehicle compartment three-dimensional point cloud data in step S102 includes: when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is less than the preset resolution (i.e., below 32×20 resolution), determining the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment three-dimensional point cloud data; when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is greater than or equal to the preset resolution, determining the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment image.
[0042] Specifically, whether the vehicle can be identified by recognizing the license plate number is determined based on the resolution of the license plate number part recognized on the vehicle compartment image. When the resolution of the license plate number part recognized on the above vehicle compartment image is less than the preset resolution, it is determined that the type of the vehicle cannot be determined by recognizing the license plate number recognized on the vehicle compartment image. When the resolution of the license plate number part recognized on the above vehicle compartment image is greater than or equal to the preset resolution, it is determined that the type of the vehicle can be determined by recognizing the license plate number recognized on the vehicle compartment image. The above-mentioned feature set data corresponding to the above-mentioned vehicle compartment image is then determined.
[0043] Step S103, when it is determined that the quality of the carriage image does not meet the requirements, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the carriage to determine whether there is a preset abnormal point in the three-dimensional point cloud data of the carriage, and the preset abnormal point is a convex or concave abnormal point;
[0044] For example, when judging whether the rivets on the car body have fallen off through images, because the rivet cap is small in size and if the color is similar to the car body, it may be difficult to make an accurate judgment using image analysis. At this time, using laser radar to scan the number of protrusions on the car body can more conveniently obtain the detection result.
[0045] In one embodiment of the present application, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the above-mentioned car, and determine whether there are preset abnormal points in the three-dimensional point cloud data of the above-mentioned car, including: using the three-dimensional point cloud data analysis algorithm to analyze the three-dimensional point cloud data of the target car, and determine whether there are the above-mentioned preset abnormal points in the three-dimensional point cloud data of the above-mentioned car, and the three-dimensional point cloud data of the above-mentioned target car is the three-dimensional point cloud data of the above-mentioned car at the same time of acquisition as the target car image.
[0046] Step S104, when it is determined that the above-mentioned preset abnormal point exists in the above-mentioned three-dimensional point cloud data of the car, the above-mentioned preset abnormal point and feature set data are processed by using neural network model technology to obtain multiple type confidences, and the type of the above-mentioned vehicle is determined to be a type corresponding to the target type confidence, and the above-mentioned target type confidence is the maximum value of all the above-mentioned type confidences.
[0047] In the above steps, the car compartment images collected by photographic equipment and the car compartment three-dimensional point cloud data collected by lidar equipment also use three-dimensional point cloud data analysis algorithms, neural network model technology and image processing algorithms. At the same time, since the specific attributes, feature sets and self-test templates of the car compartment are also taken into consideration, the targeted nature of image detection is enhanced. Compared with the existing solutions, the type of vehicle can be identified more accurately, thereby solving the problem that the existing solutions cannot identify the type of vehicle based on the tiny identification objects (≤5cm) of the vehicle with a car compartment.
[0048] It can effectively locate small targets on dynamic large-scale equipment, and map the spatial position information of these targets to specific identification areas on the image, thereby enhancing the pertinence of image detection. In addition, this method can also simplify the processing of problems that cannot be easily distinguished by image algorithms. For example, when judging whether the rivets on the carriage have fallen off through images, because the volume of the rivet cap is small and if the color is similar to that of the carriage, it may be difficult to accurately determine it using image analysis methods. At this time, using laser radar to scan the number of protrusions on the carriage can more conveniently obtain the detection results;
[0049] Similarly, when the light is poor, such as at night or on cloudy days, it may be difficult to identify whether the side door on the carriage is closed correctly. At this time, the lidar can also make accurate judgments.
[0050] This method can not only provide a more convenient basis for image analysis algorithms to make judgments when locating small targets, but can also serve as an alternative means to support target detection when problems such as insufficient lighting conditions arise, provide a basis for judgment when the image analysis algorithm fails, define the analysis area for the image analysis algorithm, and provide three-dimensional texture features as a new basis for judgment in addition to two-dimensional graphic features.
[0051] In one embodiment of the present application, in the process of using the neural network model technology to process the above-mentioned preset abnormal point and feature set data to obtain multiple types of confidence, the above-mentioned method also includes:
[0052] When it is determined that the tarpaulin is highly overloaded, the tie bars are broken, and the car baffle is abnormal, a first alarm message is generated to prompt that the tarpaulin is highly overloaded, the tie bars are broken, and the car baffle is abnormal; when it is determined that the tarpaulin is highly overloaded, the tie bars are broken, and the car baffle is normal, a second alarm message is generated to prompt that the tarpaulin is highly overloaded and the tie bars are broken; when it is determined that the tarpaulin is highly overloaded, the tie bars are not broken, and the car baffle is normal, a third alarm message is generated to prompt that the tarpaulin is highly overloaded.
[0053] The broken tie rod is judged by whether there are tie rods in a fixed position, while the excessive height of the tarpaulin is judged by the height of the compartment baffle of the compartment. If the target is artificially modified, there is a risk of excessive height and overload, which is considered to be not in line with expectations.
[0054] Specifically, by judging whether the tarpaulin is overloaded, whether the reinforcement is broken, or whether the compartment baffle is abnormal, corresponding alarm information is generated to prompt the corresponding fault, so that the staff can be aware of the vehicle fault and deal with it in time to ensure the safe operation of the vehicle in the future.
[0055] In one embodiment of the present application, the above method also includes: when it is determined that the above specific attribute is within a preset length range, storing the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data; when it is determined that the above-mentioned specific attribute is not within a preset length range (which can be between 2cm and 5cm), determining not to store the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data.
[0056] Specifically, when an image algorithm is needed to verify whether the car connection is loose, the three-dimensional model data of the car connection needs to be saved first before it can be compared and matched with the car. The coordinates are stored for easier calibration later.
[0057] For example, if you need to use an image algorithm to verify whether the rivets on the side of the car have fallen off, you need to save the three-dimensional model data of the rivets first before you can compare and match them with the car.
[0058] When you need to use image algorithms to verify whether the car connection is loose, you need to save the three-dimensional model data of the car connection first before you can compare and match it with the car.
[0059] When you need to use an image algorithm to verify whether there are cracks on the surface of the car body, you need to save the three-dimensional model data of the car body surface first before you can compare and match it with the car body.
[0060] When you need to use image algorithms to verify whether there are cracks inside the car, you need to save the three-dimensional model data of the interior of the car first before you can compare and match it with the car.
[0061] When you need to use an image algorithm to check whether there are cracks on the wheel surface, you need to save the three-dimensional model data of the wheel surface first before you can compare and match it with the car body.
[0062] When you need to use image algorithms to verify whether the brake pad thickness is too thin, you need to first save the three-dimensional model data of the brake pad thickness before you can compare and match it with the car body.
[0063] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the vehicle identification method of the present application will be described in detail below in conjunction with specific embodiments.
[0064] This embodiment relates to a specific vehicle identification method, the vehicle has a carriage, such as Figure 2 As shown, the following steps are included:
[0065] Step S1: When the vehicle compartment enters the identification area, the vehicle compartment image captured by the photographic device is acquired in real time, and the three-dimensional point cloud data of the vehicle compartment captured by the laser radar device is acquired;
[0066] Step S2: using an image quality evaluation algorithm to score the carriage images at multiple times to obtain multiple image scores; analyzing the target carriage image, and determining whether the quality of the target carriage image meets the requirements by looking at various defects to be identified in the carriage of each image and whether the carriage number area is clear (which can be identified by the algorithm). The target carriage image is the maximum value of all image scores;
[0067] Step S3: determining feature set data corresponding to the carriage image or the carriage three-dimensional point cloud data, the image processing algorithm is a plane determination algorithm, the feature set data includes at least one of the following: specific attributes of the carriage, feature set and self-checking template, the specific attributes include the length, width and height of the carriage, the feature set includes the three-dimensional shape of the front of the vehicle and the height and width of the baffle of the carriage, the self-checking template includes the inspection template of whether the tarpaulin height is overloaded, the inspection template of whether the tie bar is broken, and the inspection template of whether the carriage baffle is abnormal;
[0068] Step S4: when it is determined that the quality of the carriage image does not meet the requirements, a point cloud-based method is used for processing, that is, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the target carriage to determine whether there are preset abnormal points in the three-dimensional point cloud data of the carriage, and the three-dimensional point cloud data of the target carriage is the three-dimensional point cloud data of the carriage at the same time of acquisition as the target carriage image;
[0069] Step S5: When it is determined that there are preset abnormal points in the three-dimensional point cloud data of the carriage, the preset abnormal points and feature set data are processed using the neural network model technology to obtain multiple type confidences, and the type of the vehicle is determined to be the type corresponding to the target type confidence, and the target type confidence is the maximum value of all type confidences. When it is determined that there are no preset abnormal points in the three-dimensional point cloud data of the carriage, it is determined that the carriage has no abnormal conditions, that is, no abnormal conditions such as damage.
[0070] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0071] The embodiment of the present application also provides a vehicle identification device. It should be noted that the vehicle identification device of the embodiment of the present application can be used to execute the vehicle identification method provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0072] The vehicle has a carriage, and the vehicle identification device provided in the embodiment of the present application is introduced below.
[0073] Figure 3 FIG. 1 is a structural block diagram of a vehicle identification device provided according to an embodiment of the present application. Figure 3 As shown, the device comprises:
[0074] The acquisition unit 31 is used to acquire the image of the vehicle compartment captured by the photographic device and the three-dimensional point cloud data of the vehicle compartment captured by the laser radar device in real time when the vehicle compartment enters the recognition area;
[0075] The first determination unit 32 is used to analyze the above-mentioned carriage image by using an image processing algorithm to determine whether the quality of the above-mentioned carriage image meets the requirements, and determine the feature set data corresponding to the above-mentioned carriage image or the above-mentioned carriage three-dimensional point cloud data, the above-mentioned image processing algorithm is a plane determination algorithm, and the above-mentioned feature set data includes at least one of the following: specific attributes, feature sets and self-checking templates of the above-mentioned carriage, the above-mentioned specific attributes include the length, width and height of the above-mentioned carriage, the above-mentioned feature set includes the three-dimensional shape of the front of the vehicle and the height and width of the baffle of the above-mentioned carriage, and the above-mentioned self-checking template includes an inspection template for whether the tarpaulin height is overloaded, an inspection template for whether the tie rod is broken, and an inspection template for whether the carriage baffle is abnormal;
[0076] The second determination unit 33 is used to analyze the three-dimensional point cloud data of the carriage using a three-dimensional point cloud data analysis algorithm to determine whether there is a preset abnormal point in the three-dimensional point cloud data of the carriage when it is determined that the quality of the carriage image does not meet the requirements, and the preset abnormal point is a convex or concave abnormal point;
[0077] The processing unit 34 is used to use the neural network model technology to process the above-mentioned preset abnormal point and feature set data when it is determined that the above-mentioned preset abnormal point exists in the above-mentioned three-dimensional point cloud data of the vehicle compartment, to obtain multiple type confidences, and determine that the type of the above-mentioned vehicle is a type corresponding to the target type confidence, and the above-mentioned target type confidence is the maximum value of all the above-mentioned type confidences.
[0078] In the above-mentioned device, the car compartment image is collected by photographic equipment and the car compartment three-dimensional point cloud data is collected by lidar equipment, which also adopts a three-dimensional point cloud data analysis algorithm, a neural network model technology and an image processing algorithm. At the same time, since the specific attributes, feature sets and self-test templates of the car compartment are also taken into consideration, the targetedness of image detection is enhanced. Compared with the existing scheme, the type of vehicle can be identified more accurately, thereby solving the problem that the existing scheme cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a car compartment.
[0079] In one embodiment of the present application, the first determination unit includes a first processing module and a first determination module. The first processing module is used to use an image quality evaluation algorithm to score the above-mentioned carriage images at multiple times to obtain multiple image scores; the first determination module is used to analyze the target carriage image to determine whether the quality of the above-mentioned target carriage image meets the requirements, and the above-mentioned target carriage image is the maximum value of all the above-mentioned image scores.
[0080] In one embodiment of the present application, the second determination unit includes a second determination module, which is used to analyze the target car three-dimensional point cloud data using the above-mentioned three-dimensional point cloud data analysis algorithm to determine whether the above-mentioned preset abnormal points exist in the above-mentioned car three-dimensional point cloud data, and the above-mentioned target car three-dimensional point cloud data is the above-mentioned car three-dimensional point cloud data at the same time of acquisition as the target car image.
[0081] In one embodiment of the present application, the first processing module includes a processing sub-module, which is used to adopt the above-mentioned image quality evaluation algorithm to score the above-mentioned carriage images at multiple times based on image clarity, the number of obstructions, color brightness and exposure, and obtain multiple of the above-mentioned image scores.
[0082] In one embodiment of the present application, the processing unit includes a second processing module, a third processing module and a fourth processing module. In the process of using the neural network model technology to process the above-mentioned preset abnormal point and feature set data to obtain multiple types of confidence, the second processing module is used to generate a first alarm message when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rod is broken, and the car baffle is abnormal, to prompt that the above-mentioned tarpaulin is highly overloaded, the tie rod is broken, and the car baffle is abnormal; the third processing module is used to generate a second alarm message when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rod is broken, and the car baffle is normal, to prompt that the above-mentioned tarpaulin is highly overloaded and the tie rod is broken; the fourth processing module is used to generate a third alarm message when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rod is not broken, and the car baffle is normal, to prompt that the above-mentioned tarpaulin is highly overloaded.
[0083] In one embodiment of the present application, the first determination unit includes a third determination module and a fourth determination module. The third determination module is used to determine the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment three-dimensional point cloud data when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is less than a preset resolution; the fourth determination module is used to determine the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment image when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is greater than or equal to the preset resolution.
[0084] In one embodiment of the present application, the above-mentioned device also includes a storage unit and a third determination unit. The storage unit is used to store the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data when it is determined that the above-mentioned specific attribute is within a preset length range; the third determination unit is used to determine not to store the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data when it is determined that the above-mentioned specific attribute is not within the preset length range.
[0085] The above-mentioned vehicle identification device includes a processor and a memory, and the above-mentioned acquisition unit, the first determination unit, the second determination unit and the processing unit are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions. The above-mentioned modules are all located in the same processor; or, the above-mentioned modules are located in different processors in any combination.
[0086] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the problem that the existing solution cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a carriage is solved by adjusting the kernel parameters.
[0087] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.
[0088] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the vehicle identification method.
[0089] An embodiment of the present invention provides a processor, and the processor is used to run a program, wherein the vehicle identification method is executed when the program is run.
[0090] An embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, at least the following steps are implemented: when the vehicle compartment enters the identification area, a vehicle compartment image captured by a photographic device is acquired in real time, and three-dimensional point cloud data of the vehicle compartment is acquired by a laser radar device; an image processing algorithm is used to analyze the vehicle compartment image to determine whether the quality of the vehicle compartment image meets the requirements, and feature set data corresponding to the vehicle compartment image or the three-dimensional point cloud data of the vehicle compartment is determined, wherein the image processing algorithm is a plane determination algorithm, and the feature set data includes at least one of the following: specific attributes, feature sets, and self-inspection templates of the vehicle compartment, wherein the specific attributes include the length, width, and height of the vehicle compartment, and the feature set includes the three-dimensional shape of the vehicle head. As well as the height and width of the baffle of the above-mentioned carriage, the above-mentioned self-inspection template includes an inspection template for whether the tarpaulin height is overloaded, an inspection template for whether the reinforcement is broken, and an inspection template for whether the carriage baffle is abnormal; when it is determined that the quality of the above-mentioned carriage image does not meet the requirements, a three-dimensional point cloud data analysis algorithm is used to analyze the above-mentioned carriage three-dimensional point cloud data to determine whether there are preset abnormal points in the above-mentioned carriage three-dimensional point cloud data, and the above-mentioned preset abnormal points are convex or concave abnormal points; when it is determined that the above-mentioned preset abnormal points exist in the above-mentioned carriage three-dimensional point cloud data, a neural network model technology is used to process the above-mentioned preset abnormal points and feature set data to obtain multiple type confidences, and determine that the type of the above-mentioned vehicle is a type corresponding to the target type confidence, and the above-mentioned target type confidence is the maximum value of all the above-mentioned type confidences. The device in this article can be a server, PC, PAD, mobile phone, etc.
[0091] Optionally, an image processing algorithm is used to analyze the above-mentioned carriage image to determine whether the quality of the above-mentioned carriage image meets the requirements, including: using an image quality evaluation algorithm to score the above-mentioned carriage images at multiple moments to obtain multiple image scores; analyzing the target carriage image to determine whether the quality of the above-mentioned target carriage image meets the requirements, the above-mentioned target carriage image is the maximum value of all the above-mentioned image scores.
[0092] Optionally, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the above-mentioned car to determine whether there are preset abnormal points in the three-dimensional point cloud data of the above-mentioned car, including: using the three-dimensional point cloud data analysis algorithm to analyze the three-dimensional point cloud data of the target car to determine whether there are the above-mentioned preset abnormal points in the three-dimensional point cloud data of the above-mentioned car, and the three-dimensional point cloud data of the above-mentioned target car is the three-dimensional point cloud data of the above-mentioned car at the same time of acquisition as the target car image.
[0093] Optionally, an image quality evaluation algorithm is used to score the above-mentioned carriage images at multiple moments to obtain multiple image scores, including: using the above-mentioned image quality evaluation algorithm to score the above-mentioned carriage images at multiple moments based on image clarity, the number of obstructions, color brightness and exposure to obtain multiple image scores.
[0094] Optionally, in the process of using neural network model technology to process the above-mentioned preset abnormal points and feature set data to obtain multiple types of confidence, the above-mentioned method also includes: when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is abnormal, a first alarm message is generated to prompt that the above-mentioned tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is abnormal; when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is normal, a second alarm message is generated to prompt that the above-mentioned tarpaulin is highly overloaded and the tie rods are broken; when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rods are not broken, and the car baffle is normal, a third alarm message is generated to prompt that the above-mentioned tarpaulin is highly overloaded.
[0095] Optionally, determining the feature set data corresponding to the above-mentioned vehicle compartment image or the above-mentioned vehicle compartment three-dimensional point cloud data includes: when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is less than a preset resolution, determining the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment three-dimensional point cloud data; when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is greater than or equal to the preset resolution, determining the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment image.
[0096] Optionally, the above method also includes: when it is determined that the above specific attribute is within a preset length range, storing the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data; when it is determined that the above specific attribute is not within the preset length range, determining not to store the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data.
[0097] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that is initialized with at least the following method steps: when the cabin of the above-mentioned vehicle enters the identification area, obtaining in real time a cabin image captured by a photographic device, and obtaining three-dimensional point cloud data of the cabin captured by a lidar device; using an image processing algorithm to analyze the above-mentioned cabin image to determine whether the quality of the above-mentioned cabin image meets the requirements, and determining feature set data corresponding to the above-mentioned cabin image or the above-mentioned cabin three-dimensional point cloud data, the above-mentioned image processing algorithm is a plane judgment algorithm, and the above-mentioned feature set data includes at least one of the following: specific attributes, feature sets and self-inspection templates of the above-mentioned cabin, the above-mentioned specific attributes include the length, width and height of the above-mentioned cabin, and the above-mentioned feature set includes the three-dimensional shape of the front of the vehicle and the above-mentioned cabin the height and width of the baffle, the self-inspection template includes an inspection template for whether the tarpaulin height is overloaded, an inspection template for whether the reinforcement is broken, and an inspection template for whether the car baffle is abnormal; when it is determined that the quality of the above-mentioned car image does not meet the requirements, a three-dimensional point cloud data analysis algorithm is used to analyze the above-mentioned car three-dimensional point cloud data to determine whether there are preset abnormal points in the above-mentioned car three-dimensional point cloud data, and the above-mentioned preset abnormal points are raised or recessed abnormal points; when it is determined that the above-mentioned preset abnormal points exist in the above-mentioned car three-dimensional point cloud data, a neural network model technology is used to process the above-mentioned preset abnormal points and feature set data to obtain multiple type confidences, and the type of the above-mentioned vehicle is determined to be a type corresponding to the target type confidence, and the above-mentioned target type confidence is the maximum value of all the above-mentioned type confidences.
[0098] Optionally, an image processing algorithm is used to analyze the above-mentioned carriage image to determine whether the quality of the above-mentioned carriage image meets the requirements, including: using an image quality evaluation algorithm to score the above-mentioned carriage images at multiple moments to obtain multiple image scores; analyzing the target carriage image to determine whether the quality of the above-mentioned target carriage image meets the requirements, the above-mentioned target carriage image is the maximum value of all the above-mentioned image scores.
[0099] Optionally, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the above-mentioned car to determine whether there are preset abnormal points in the three-dimensional point cloud data of the above-mentioned car, including: using the three-dimensional point cloud data analysis algorithm to analyze the three-dimensional point cloud data of the target car to determine whether there are the above-mentioned preset abnormal points in the three-dimensional point cloud data of the above-mentioned car, and the three-dimensional point cloud data of the above-mentioned target car is the three-dimensional point cloud data of the above-mentioned car at the same time of acquisition as the target car image.
[0100] Optionally, an image quality evaluation algorithm is used to score the above-mentioned carriage images at multiple moments to obtain multiple image scores, including: using the above-mentioned image quality evaluation algorithm to score the above-mentioned carriage images at multiple moments based on image clarity, the number of obstructions, color brightness and exposure to obtain multiple image scores.
[0101] Optionally, in the process of using neural network model technology to process the above-mentioned preset abnormal points and feature set data to obtain multiple types of confidence, the above-mentioned method also includes: when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is abnormal, a first alarm message is generated to prompt that the above-mentioned tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is abnormal; when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rods are broken, and the car baffle is normal, a second alarm message is generated to prompt that the above-mentioned tarpaulin is highly overloaded and the tie rods are broken; when it is determined that the above-mentioned tarpaulin is highly overloaded, the tie rods are not broken, and the car baffle is normal, a third alarm message is generated to prompt that the above-mentioned tarpaulin is highly overloaded.
[0102] Optionally, determining the feature set data corresponding to the above-mentioned vehicle compartment image or the above-mentioned vehicle compartment three-dimensional point cloud data includes: when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is less than a preset resolution, determining the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment three-dimensional point cloud data; when the resolution of the license plate number part recognized on the above-mentioned vehicle compartment image is greater than or equal to the preset resolution, determining the above-mentioned feature set data corresponding to the above-mentioned vehicle compartment image.
[0103] Optionally, the above method also includes: when it is determined that the above specific attribute is within a preset length range, storing the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data; when it is determined that the above specific attribute is not within the preset length range, determining not to store the coordinates of each point cloud of the above-mentioned car three-dimensional point cloud data.
[0104] The present application also provides a vehicle identification system, characterized in that it includes: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include methods for executing any of the above-mentioned vehicle identification methods. The car compartment image collected by the photographic device and the three-dimensional point cloud data of the car compartment collected by the laser radar device, wherein the three-dimensional point cloud data analysis algorithm, neural network model technology and image processing algorithm are also used, and at the same time, because the specific attributes, feature sets and self-test templates of the car compartment are also added as considerations, the pertinence of the image detection is enhanced, and the type of vehicle can be identified more accurately than the existing solution, thereby solving the problem that the existing solution cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a car compartment.
[0105] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.
[0106] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0107] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0108] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0110] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0111] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0112] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0113] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0114] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:
[0115] 1) The vehicle identification method of the present application uses the car compartment image collected by photographic equipment and the car compartment three-dimensional point cloud data collected by lidar equipment, which also adopts a three-dimensional point cloud data analysis algorithm, a neural network model technology and an image processing algorithm. At the same time, since the specific attributes, feature sets and self-test templates of the car compartment are also taken into consideration, the targetedness of image detection is enhanced. Compared with the existing solutions, the type of vehicle can be identified more accurately, thereby solving the problem that the existing solutions cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a car compartment.
[0116] 2) The vehicle identification device of the present application uses the car compartment image collected by the photographic equipment and the car compartment three-dimensional point cloud data collected by the lidar equipment, and also adopts the three-dimensional point cloud data analysis algorithm, the neural network model technology and the image processing algorithm. At the same time, because the specific attributes, feature sets and self-test templates of the car compartment are also taken into consideration, the targetedness of the image detection is enhanced. Compared with the existing scheme, the type of vehicle can be identified more accurately, thereby solving the problem that the existing scheme cannot identify the type of vehicle based on the tiny identification object (≤5cm) of the vehicle with a car compartment.
[0117] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A method for identifying a vehicle, wherein the vehicle has a carriage, characterized in that: include: When the vehicle compartment enters the identification area, a vehicle compartment image captured by a photographic device is acquired in real time, and three-dimensional point cloud data of the vehicle compartment is acquired by a laser radar device; An image processing algorithm is used to analyze the carriage image to determine whether the quality of the carriage image meets the requirements, and feature set data corresponding to the carriage image or the three-dimensional point cloud data of the carriage is determined. The image processing algorithm is a plane determination algorithm. The feature set data includes at least one of the following: specific attributes of the carriage, feature set and self-check template. The specific attributes include the length, width and height of the carriage. The feature set includes the three-dimensional shape of the front of the vehicle and the height and width of the baffle of the carriage. The self-check template includes an inspection template for whether the tarpaulin is overloaded, an inspection template for whether the reinforcement is broken, and an inspection template for whether the baffle of the carriage is abnormal. When it is determined that the quality of the carriage image does not meet the requirements, a three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the carriage to determine whether there is a preset abnormal point in the three-dimensional point cloud data of the carriage, and the preset abnormal point is a convex or concave abnormal point; When it is determined that the preset abnormal point exists in the three-dimensional point cloud data of the vehicle compartment, the neural network model technology is used to process the preset abnormal point and feature set data to obtain multiple type confidences, and the type of the vehicle is determined to be a type corresponding to the target type confidence, and the target type confidence is the maximum value of all the type confidences.
2. The method according to claim 1, characterized in that: An image processing algorithm is used to analyze the carriage image to determine whether the quality of the carriage image meets the requirements, including: Using an image quality evaluation algorithm to score the carriage images at multiple moments to obtain multiple image scores; The target compartment image is analyzed to determine whether the quality of the target compartment image meets the requirements, and the target compartment image is the maximum value of all the image scores.
3. The method according to claim 2, characterized in that The three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the carriage to determine whether there are preset abnormal points in the three-dimensional point cloud data of the carriage, including: The three-dimensional point cloud data analysis algorithm is used to analyze the three-dimensional point cloud data of the target compartment to determine whether the preset abnormal point exists in the three-dimensional point cloud data of the compartment, and the three-dimensional point cloud data of the target compartment is the three-dimensional point cloud data of the compartment at the same time of acquisition as the target compartment image.
4. The method according to claim 2, characterized in that: An image quality evaluation algorithm is used to score the carriage images at multiple moments to obtain multiple image scores, including: The image quality evaluation algorithm is used to score the carriage images at multiple moments based on image clarity, the number of obstructions, color brightness and exposure, to obtain multiple image scores.
5. The method according to claim 1, characterized in that In the process of using the neural network model technology to process the preset abnormal point and feature set data to obtain multiple types of confidence, the method also includes: When it is determined that the tarpaulin is highly overloaded, the tie bars are broken, and the compartment baffle is abnormal, a first alarm message is generated to prompt that the tarpaulin is highly overloaded, the tie bars are broken, and the compartment baffle is abnormal; When it is determined that the tarpaulin is highly overloaded and the tie bars are broken, and the compartment baffle is normal, a second alarm message is generated to prompt that the tarpaulin is highly overloaded and the tie bars are broken; When it is determined that the tarpaulin is highly overloaded, the tie bars are not broken, and the compartment baffle is normal, a third alarm message is generated to prompt that the tarpaulin is highly overloaded.
6. The method according to claim 1, characterized in that Determining feature set data corresponding to the carriage image or the carriage three-dimensional point cloud data includes: When the resolution of the license plate number portion recognized on the carriage image is less than a preset resolution, determining the feature set data corresponding to the three-dimensional point cloud data of the carriage; When the resolution of the license plate number portion recognized on the vehicle compartment image is greater than or equal to a preset resolution, the feature set data corresponding to the vehicle compartment image is determined.
7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: When it is determined that the specific attribute is within a preset length range, storing the coordinates of each point cloud of the three-dimensional point cloud data of the carriage; When it is determined that the specific attribute is not within the preset length range, it is determined not to store the coordinates of each point cloud of the three-dimensional point cloud data of the carriage.
8. A vehicle identification device, the vehicle having a carriage, characterized in that: include: an acquisition unit, used to acquire, in real time, an image of the vehicle compartment acquired by a photographic device and three-dimensional point cloud data of the vehicle compartment acquired by a laser radar device when the vehicle compartment enters a recognition area; A first determination unit is used to analyze the carriage image by using an image processing algorithm to determine whether the quality of the carriage image meets the requirements, and determine feature set data corresponding to the carriage image or the three-dimensional point cloud data of the carriage, wherein the image processing algorithm is a plane determination algorithm, and the feature set data includes at least one of the following: specific attributes of the carriage, a feature set, and a self-checking template, wherein the specific attributes include the length, width, and height of the carriage, the feature set includes the three-dimensional shape of the front of the vehicle and the height and width of the baffle of the carriage, and the self-checking template includes an inspection template for whether the tarpaulin height is overloaded, an inspection template for whether the tie rod is broken, and an inspection template for whether the carriage baffle is abnormal; A second determination unit is used to analyze the three-dimensional point cloud data of the carriage using a three-dimensional point cloud data analysis algorithm when it is determined that the quality of the carriage image does not meet the requirements, and determine whether there is a preset abnormal point in the three-dimensional point cloud data of the carriage, wherein the preset abnormal point is a convex or concave abnormal point; A processing unit is used to, when it is determined that the preset abnormal point exists in the three-dimensional point cloud data of the vehicle compartment, use a neural network model technology to process the preset abnormal point and feature set data to obtain multiple type confidences, and determine that the type of the vehicle is a type corresponding to a target type confidence, wherein the target type confidence is the maximum value of all the type confidences.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the vehicle identification method according to any one of claims 1 to 7.
10. A vehicle identification system, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the vehicle identification method described in any one of claims 1 to 7.
