Vehicle track image and track coordinate processing method and device, equipment and medium

Abstract

The invention provides a vehicle track image and track coordinate processing method and device, equipment and a medium, and the method comprises the steps: obtaining a first track image of a target vehicle, and carrying out the preprocessing of the first track image; performing reconstruction processing on the preprocessed first track image by using a variational auto-encoder model to obtain a second track image; sequentially performing quality evaluation, color space conversion and trajectory contour extraction on the second trajectory image to obtain a trajectory contour, and performing coordinate identification and coordinate extraction on the trajectory contour to obtain coordinate points of the trajectory contour; post-processing the coordinate points to obtain an original coordinate point set of the target vehicle, and obtaining a track coordinate processing result according to the original coordinate point set. The method has the beneficial effect that the automatic generation efficiency and accuracy of the vehicle track are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automatic driving and intelligent transportation, and in particular to a vehicle trajectory image and trajectory coordinate processing method, device, equipment and medium. BACKGROUND

[0002] In the context of automatic driving and intelligent transportation systems, vehicle trajectory prediction and analysis are key technologies to ensure driving safety and improve road efficiency. However, existing vehicle trajectory data acquisition methods have some limitations, especially in handling complex traffic scenarios and variable driving behaviors. In order to achieve more accurate and reliable vehicle trajectory prediction, it is necessary to deeply study vehicle trajectory generation and analysis methods.

[0003] Currently, vehicle trajectory generation methods are mainly divided into physical model-based, probabilistic model-based and data-driven methods. Physical model-based methods predict trajectories by simulating the dynamics equations of vehicle motion, but such methods often ignore the complexity of traffic environment and vehicle interaction. Probabilistic models, such as Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM), can learn the probability distribution of vehicle behavior from historical data, but they are usually limited by model assumptions and difficult to capture non-linear and high-dimensional data features. Data-driven methods, especially deep learning-based methods such as Recurrent Neural Network (RNN) and Long Short-Term Memory Network (LSTM), perform well in handling large-scale data sets and complex pattern recognition, but they usually require a large amount of labeled data and are highly sensitive to input features.

[0004] Existing vehicle trajectory image generation methods, such as those based on Generative Adversarial Networks (GAN), can generate realistic trajectory images, but they often focus on the visual quality of images while ignoring the physical reasonableness and prediction accuracy of trajectory data. In addition, existing methods mainly rely on manual labeling or simple image processing techniques for trajectory coordinate data extraction, which is inefficient in handling large-scale data sets or complex scenarios and is prone to introducing human errors. SUMMARY

[0005] The main purpose of the embodiments of the present application is to provide a vehicle trajectory image and trajectory coordinate processing method, device, equipment and medium to improve the efficiency and accuracy of automatic generation of vehicle trajectories.

[0006] One aspect of the present application provides a vehicle trajectory image and trajectory coordinate processing method, comprising:

[0007] obtaining a first trajectory image of a target vehicle, and performing preprocessing on the first trajectory image;

[0008] reconstructing the preprocessed first trajectory image using a variational autoencoder model to obtain a second trajectory image;

[0009] performing quality evaluation, color space conversion and trajectory contour extraction on the second trajectory image in sequence to obtain a trajectory contour, performing coordinate identification and coordinate extraction on the trajectory contour to obtain coordinate points of the trajectory contour;

[0010] performing post-processing on the coordinate points to obtain an original coordinate point set of the target vehicle, and obtaining a trajectory coordinate processing result according to the original coordinate point set.

[0011] According to the vehicle trajectory image and trajectory coordinate processing method, the preprocessing includes at least one of gray scale, normalized size and data enhancement.

[0012] According to the vehicle trajectory image and trajectory coordinate processing method, the training of the variational autoencoder model includes:

[0013] The preprocessed data set is taken as input, iterative updating of model parameters is performed according to a loss function and an Adam optimizer, model training is performed, and a variational autoencoder model is obtained.

[0014] According to the vehicle trajectory image and trajectory coordinate processing method, the loss function includes a combination of reconstruction loss, KL divergence loss and color loss.

[0015] The loss function is:

[0016]

[0017] wherein, is the loss function, is the reconstruction loss, is the KL divergence loss, is the color loss, and α is a weight parameter for adjusting the KL divergence loss and β is a parameter for adjusting the color loss.

[0018] The reconstruction loss is used to calculate the binary cross entropy between the second trajectory image and the first trajectory image.

[0019] The KL divergence loss is the difference between the latent variable distribution and the prior distribution, and the expression is:

[0020]

[0021] wherein, μ i represents the mean of the i th latent variable, and σ i represents the variance of the i th latent variable.

[0022] wherein, wherein the color loss is a measure of the difference between the color channel in the second trajectory image and the color channel in the first trajectory image.

[0023] The vehicle trajectory image and trajectory coordinate processing method, wherein the first trajectory image after preprocessing is reconstructed by using a variational autoencoder model to obtain a second trajectory image, comprising:

[0024] The variational autoencoder model comprises an encoder and a decoder, and the first trajectory image is mapped to a latent space by the encoder;

[0025] The latent vector is sampled in the latent space by a reparameterization method;

[0026] The latent vector is reconstructed by the decoder to obtain the second trajectory image, wherein the encoder comprises three linear layers, one of which is used to flatten and map the input image to an intermediate dimensional space, and the other two are used to output the mean and log variance of the latent representation respectively; the decoder comprises two linear layers, and the decoder maps back from the latent space to the data space and reconstructs the image by the two linear layers.

[0027] The vehicle trajectory image and trajectory coordinate processing method, wherein the second trajectory image is sequentially subjected to quality evaluation, color space conversion and trajectory contour extraction to obtain a trajectory contour, and the trajectory contour is subjected to coordinate identification and coordinate extraction to obtain coordinate points of the trajectory contour, comprising:

[0028] The second trajectory image is evaluated in quality by using an MSE and PSNR index evaluation method, wherein the quality evaluation is used to determine the authenticity of the second trajectory image;

[0029] The second trajectory image is converted from an RGB color space to an HSV color space;

[0030] A mask is created in the HSV color space according to a preset color range, and the extraction of the trajectory contour of different colors is performed according to the mask;

[0031] The coordinate identification and coordinate extraction are performed by using an image processing algorithm to obtain the coordinate points of the trajectory contour.

[0032] The vehicle trajectory image and trajectory coordinate processing method, wherein the coordinate points are post-processed to obtain a set of original coordinate points of the target vehicle, and a trajectory coordinate processing result is obtained according to the set of original coordinate points, comprising:

[0033] The post-processing comprises denoising and normalization processing, wherein the normalized x norm is:

[0034]

[0035] wherein X is the set of extracted original coordinate points, and x represents the coordinate point data in the set of original coordinate points X;

[0036] Save the original coordinate point set obtained after post-processing to a preset format file as coordinate data to obtain a trajectory coordinate processing result.

[0037] Another aspect of the embodiment of the application provides a vehicle trajectory image and trajectory coordinate processing device, comprising:

[0038] The first module is configured to acquire a first trajectory image of a target vehicle, and perform preprocessing on the first trajectory image.

[0039] The second module is configured to perform reconstruction processing on the preprocessed first trajectory image by using a variational autoencoder model to obtain a second trajectory image.

[0040] The third module is configured to sequentially perform quality evaluation, color space conversion and trajectory contour extraction on the second trajectory image to obtain a trajectory contour, and perform coordinate identification and coordinate extraction on the trajectory contour to obtain coordinate points of the trajectory contour.

[0041] The fourth module is configured to perform post-processing on the coordinate points to obtain an original coordinate point set of the target vehicle, and obtain a trajectory coordinate processing result according to the original coordinate point set.

[0042] Another aspect of the embodiment of the application provides an electronic device comprising a processor and a memory.

[0043] The memory is configured to store a program.

[0044] The processor executes the program to implement the method as described above.

[0045] The embodiment of the application also discloses a computer program product or a computer program, which comprises computer instructions stored in a computer readable storage medium. A processor of a computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to enable the computer device to execute the method as described above.

[0046] The embodiment of the application has the following beneficial effects: the variational autoencoder learns the potential distribution of vehicle trajectory data, can generate synthetic trajectory images conforming to physical characteristics and driving behavior patterns, and adopts an iterative correction strategy to optimize the sampling process of the potential space, thereby improving the diversity and authenticity of the generated images; the automatic trajectory coordinate data extraction method is adopted, the trajectory coordinate data can be accurately and quickly extracted from the generated images by combining color space conversion and image processing, and the efficiency and accuracy of trajectory data processing are improved. BRIEF DESCRIPTION OF DRAWINGS

[0047] The above and / or additional aspects and advantages of the present application will become apparent and more readily appreciated from the following description, taken in conjunction with the following drawings of which:

[0048] Figure 1 is a vehicle trajectory image and trajectory coordinate processing flowchart of an embodiment of the present application.

[0049] Figure 2 is a variational autoencoder model schematic diagram of an embodiment of the present application.

[0050] Figure 3 is a trajectory image reconstruction flowchart of an embodiment of the present application.

[0051] Figure 4 is a collision scene graph contained in an original data set of an embodiment of the present application.

[0052] Figure 5 is a trajectory coordinate processing flowchart of an embodiment of the present application.

[0053] Figure 6 is a trajectory contour extraction schematic diagram of an embodiment of the present application.

[0054] Figure 7 is a new vehicle trajectory picture generated from an original scene of an embodiment of the present application.

[0055] Figure 8 is a vehicle trajectory image and trajectory coordinate processing device schematic diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0056] Embodiments of the present application are described in detail below with reference to the attached drawings, which show by way of example, embodiments in which the same or similar elements are denoted by the same or similar reference numerals throughout the various figures. In the following description, the suffixes "module", "part" or "unit" used to denote elements are merely intended for facilitating explanation of the present application, and do not have their own particular meaning or function. Therefore, "module", "part" or "unit" can be used interchangeably. "First", "second", and so on are used only to distinguish technical features for the purpose of explanation, and cannot be understood to indicate or imply relative importance or to imply the number of technical features indicated or the order of technical features indicated. In the following description, consecutive numbers of method steps are for the convenience of review and understanding, and adjusting the order of implementation between steps does not affect the technical effects achieved by the overall technical solution of the present application in combination with the logical relationship between the steps. The embodiments described below with reference to the drawings are exemplary and are only used to explain the present application, and cannot be understood as limiting the present application.

[0057] Figure 1is a vehicle trajectory image and trajectory coordinate processing flowchart of an embodiment of the present application, which includes but is not limited to steps S100-S400:

[0058] S100, obtaining a first trajectory image of a target vehicle, and performing preprocessing on the first trajectory image.

[0059] The first trajectory image is an original trajectory image of the target vehicle.

[0060] In some embodiments, the collected vehicle trajectory image is preprocessed, including grayscale, normalized size, and possible data enhancement steps, to improve the generalization ability of the model to different condition images.

[0061] S200, reconstructing the preprocessed first trajectory image using a variational autoencoder model to obtain a second trajectory image.

[0062] The second trajectory image is a new image obtained after reconstruction.

[0063] In some embodiments, the second trajectory image is obtained by Figure 2 wherein Figure 2 is a schematic diagram of a variational autoencoder model. The variational autoencoder model is obtained by training, and the variational autoencoder model is used to reconstruct the original trajectory image to obtain the latent representation of the original trajectory image and generate diversified trajectory images.

[0064] In some embodiments, the training of the variational autoencoder model includes: taking the preprocessed data set as input, iteratively updating the model parameters according to the loss function and the Adam optimizer, training the model, and obtaining the variational autoencoder model. The model includes two main components: an encoder and a decoder. The encoder is responsible for converting the input image into a low-dimensional representation in the latent space, while the decoder converts these latent representations back into the image space to generate new vehicle trajectory images. In the latent space, the mean and log variance of the latent variables obtained by the encoder are used to sample using the reparameterization trick, thereby generating diversified trajectory images.

[0065] In some embodiments, the loss function includes a combination of reconstruction loss, KL divergence loss, and color loss.

[0066] The loss function is:

[0067]

[0068] wherein, is the loss function, is the reconstruction loss, is the KL divergence loss, for color loss, a is a weight parameter to adjust the KL divergence loss, and b is a parameter to adjust the color loss;

[0069] wherein the reconstruction loss is used to measure the binary cross-entropy between the second trajectory image and the first trajectory image;

[0070] The KL divergence loss is the difference between the latent variable distribution and the prior distribution, and the expression is:

[0071]

[0072] wherein μ i represents the mean of the i-th latent variable, and σ i represents the variance of the i-th latent variable.

[0073] wherein wherein the color loss is a measure of the difference between the color channel in the second trajectory image and the color channel in the first trajectory image.

[0074] It can be understood that the loss function is used to ensure that the generated trajectory image is visually similar to the real trajectory image and maintains physical rationality.

[0075] wherein the second trajectory image is reconstructed by the decoder from the latent vector, wherein the encoder includes three linear layers, one of which is used to flatten and map the input image to an intermediate dimensional space, and the other two are used to output the mean and logarithmic variance of the latent representation, respectively; the decoder includes two linear layers, and the decoder maps back from the latent space to the data space and reconstructs the image through the two linear layers.

[0076] In some embodiments, the reference is made to Figure 3 The trajectory image reconstruction flowchart shown includes but is not limited to steps S210-S230:

[0077] S210, the variational autoencoder model includes an encoder and a decoder, and the first trajectory image is mapped to the latent space by the encoder;

[0078] S220, the latent vector is sampled in the latent space by the reparameterization method;

[0079] S230, the second trajectory image is reconstructed by the decoder from the latent vector.

[0080] In some embodiments, the reference is made to Figure 4 The new vehicle trajectory image generated by the original scene is shown, and a large number of realistic pre-collision trajectory images are generated by the trained variational autoencoder model, which are used for subsequent automatic driving vehicle safety analysis and collision warning system development.

[0081] S300, sequentially performing quality evaluation, color space conversion and trajectory contour extraction on the second trajectory image to obtain a trajectory contour, performing coordinate recognition and coordinate extraction on the trajectory contour to obtain coordinate points of the trajectory contour.

[0082] In some embodiments, referring to FIG. 3, a flowchart of a trajectory coordinate processing method is shown, which includes but is not limited to steps S310-S340. Figure 5

[0083] S310, using the MSE and PSNR index evaluation method to evaluate the quality of the second trajectory image, wherein the quality evaluation is used to determine the authenticity of the second trajectory image.

[0084] S320, converting the second trajectory image from the RGB color space to the HSV color space.

[0085] In some embodiments, referring to FIG. 3, a flowchart of a trajectory coordinate processing method is shown, which includes but is not limited to steps S310-S340. Figure 6 In some embodiments, referring to FIG. 4, a schematic diagram of the extracted trajectory contour is shown, which includes red and blue trajectory contours.

[0086] S330, creating a mask in the HSV color space according to a preset color range, and extracting trajectory contours of different colors according to the mask.

[0087] S340, performing coordinate recognition and coordinate extraction by using an image processing algorithm to obtain coordinate points of the trajectory contour.

[0088] S400, post-processing the coordinate points to obtain a set of original coordinate points of the target vehicle, and obtaining a trajectory coordinate processing result according to the set of original coordinate points.

[0089] In some embodiments, using image processing technology, red and blue trajectories in the image are recognized and extracted; for each color trajectory, contour coordinate points are extracted and normalized; the normalized coordinate points are filled to a fixed length to form a complete trajectory data sequence.

[0090] In some embodiments, the coordinate data post-processing includes denoising and normalization processing of the extracted coordinate points, and the formula wherein X is a set of original coordinate points, and x represents coordinate point data in the set of original coordinate points X.

[0091] In some embodiments, the processed coordinate data is saved to a CSV file, and the coordinates of each trajectory point are saved with a precision of six decimal places.

[0092] In some embodiments, the experimental process of the vehicle trajectory image and the trajectory coordinate processing is as follows:

[0093] ​The experimental environment is configured, and the experiment is performed on the Autodl cloud platform using PyTorch 1.10.0 and Python 3.8 with NVIDIA RTX 4090 GPU.

[0094] Dataset preparation: A dataset of 595 images is prepared, including frontal collisions, side collisions, and rear-end collisions, as shown in Figure 4 .

[0095] Model training: Load the training dataset, instantiate the VAE model, set the intermediate dimension to 400 and the latent dimension to 20, and perform the training process.

[0096] Trajectory image generation: Use the trained model to generate 100 trajectory samples for each collision scenario, some generated images are shown in Figure 7 , a total of 59,500 samples, and quality evaluation is performed as shown in Table 1.

[0097] Coordinate extraction and processing: Apply HSV color space conversion to the generated trajectory images, extract the contours of red and blue trajectories, and extract coordinate points, as shown in Figure 6 .

[0098] Data saving: Save the extracted coordinate data to a CSV file, and some trajectory coordinate data is shown in Table 2.

[0099] Table 1 Quality evaluation table of generated images

[0100] Evaluation index Numerical value MSE 0.0271 PSNR 16.336 RMSE 0.1586

[0101] Table 2 Partial data of vehicle trajectory coordinates generated in a certain scenario

[0102] CX1 CY1 CX2 CY2 8.375514592 2.482029513 -3.309293566 1.037470455 8.522453444 2.482029513 -3.309293566 1.210382198 3.600001886 2.482029513 -3.288991151 1.383293941 3.52653246 2.606130988 -3.288991151 1.556205683 3.453063033 2.606130988 -3.268688737 1.729117426 3.379593607 2.730232464 -3.268688737 2.074940911 3.159185328 2.730232464 -3.248386322 2.247852653 4.040818443 2.730232464 -3.248386322 2.420764396 4.11428787 2.606130988 -3.228083908 2.593676139

[0103] Figure 8 is a schematic diagram of a vehicle trajectory image and trajectory coordinate processing device according to an embodiment of the present application. The device includes a first module 810, a second module 820, a third module 830, and a fourth module 840.

[0104] The first module is configured to obtain a first trajectory image of a target vehicle and perform preprocessing on the first trajectory image. The second module is configured to perform reconstruction processing on the preprocessed first trajectory image using a variational autoencoder model to obtain a second trajectory image. The third module is configured to sequentially perform quality evaluation, color space conversion, and trajectory contour extraction on the second trajectory image to obtain a trajectory contour, and perform coordinate identification and coordinate extraction on the trajectory contour to obtain coordinate points of the trajectory contour. The fourth module is configured to perform post-processing on the coordinate points to obtain a set of original coordinate points of the target vehicle, and obtain a trajectory coordinate processing result based on the set of original coordinate points.

[0105] Exemplarily, under the cooperation of the first module, the second module, the third module and the fourth module in the device, the embodiment device can realize any one of the preceding vehicle trajectory image and trajectory coordinate processing methods, that is, obtaining a first trajectory image of a target vehicle, performing preprocessing on the first trajectory image; performing reconstruction processing on the preprocessed first trajectory image by using a variational autoencoder model to obtain a second trajectory image; sequentially performing quality evaluation, color space conversion and trajectory contour extraction on the second trajectory image to obtain a trajectory contour, performing coordinate identification and coordinate extraction on the trajectory contour to obtain coordinate points of the trajectory contour; performing post-processing on the coordinate points to obtain an original coordinate point set of the target vehicle, and obtaining a trajectory coordinate processing result according to the original coordinate point set. The beneficial effects of the present application are that: the variational autoencoder learns the latent distribution of vehicle trajectory data, can generate synthetic trajectory images conforming to physical characteristics and driving behavior patterns, and uses an iterative correction strategy to optimize the sampling process of the latent space, thereby improving the diversity and authenticity of the generated images; the automatic trajectory coordinate data extraction method can accurately and quickly extract trajectory coordinate data from the generated images by combining color space conversion and image processing, thereby improving the efficiency and accuracy of trajectory data processing.

[0106] The embodiment of the present application also provides an electronic device, which comprises a processor and a memory;

[0107] The memory stores a program;

[0108] The processor executes the program to perform the vehicle trajectory image and trajectory coordinate processing method described above; the electronic device has the function of carrying and running the software system for vehicle trajectory image and trajectory coordinate processing provided by the embodiment of the present application, for example, a personal computer, a mini computer, a mainframe, a workstation, a network or a distributed computing environment, a separate or integrated computer platform, or communication with a charged particle tool or other imaging device, etc.

[0109] The embodiment of the present application also provides a computer readable storage medium, which stores a program, and the program is executed by a processor to realize the vehicle trajectory image and trajectory coordinate processing method as described above.

[0110] In some alternative embodiments, the functions / operations described in the block diagrams can not occur in the order presented in the operational illustrations. For example, two blocks shown in succession can in fact be executed substantially concurrently or the blocks can sometimes be executed in the reverse order, depending upon the functionality / operations involved. Also, although the embodiments presented in the flow diagrams are shown as a sequence of operations, it is to be understood that the logical flow is merely illustrative of alternative embodiments. The disclosed methods are not limited to the order of operations presented herein. Alternative embodiments can be conceived in which the order of operations is changed and in which sub-operations described as part of a larger operation are independently executed.

[0111] The embodiments of the present application further disclose a computer program product or computer program, which comprises computer instructions stored in a computer readable storage medium. A processor of a computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes the aforementioned vehicle trajectory image and trajectory coordinate processing method.

[0112] Furthermore, although the present application is described in the context of functional modules, it is to be understood that one or more of the described functions and / or features can be integrated in a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It is also to be understood that detailed discussion of the actual implementation of each module is unnecessary to an understanding of the present application. Rather, the actual implementation is within the routine of an engineer's knowledge given the property, functionality and internal relationships of the various functional modules disclosed herein. Accordingly, the present application is not limited to purely hardware or software implementations, but rather encompasses hybrid implementations wherein all or some of the functions are implemented in hardware and / or software. It is also to be understood that the disclosed particular concepts are merely illustrative and not intended to limit the scope of the present application, which is defined by the full scope of the appended claims and equivalents thereof.

[0113] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the technical solutions that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0114] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, which can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch the instructions from the instruction execution system, apparatus or device and execute the instructions, or in conjunction with these instructions execution systems, apparatus or devices. For the purpose of this specification, the "computer-readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by or in connection with an instruction execution system, apparatus or device, or in conjunction with these instruction execution systems, apparatus or devices.

[0115] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection having one or more wires (electrical devices), a portable computer diskette (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or other suitable medium on which the program can be printed, because the program can be electronically obtained, for example, by optical scanning of the paper or other medium, followed by editing, interpreting or otherwise processing, if necessary, in other suitable ways, to be electronically obtained and then stored in the computer memory.

[0116] It should be understood that aspects of the application can be implemented in hardware, software, firmware or a combination thereof. In the above embodiments, various steps or methods can be implemented in software or firmware which is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, and in another embodiment, any of the following technologies, known in the art, can be used: a combination of discrete logic circuits having logic gates for implementing logic functions upon an application of data signals, application specific integrated circuits having logic gates, field programmable gate arrays (FPGA), or other components, in combination or as the case can be.

[0117] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the application. In the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in one or more embodiments or examples.

[0118] Although the embodiments of the present application have been shown and described, it would be appreciated by those skilled in the art that changes, modifications, alternatives and variations can be made to the embodiments without departing from the principles and spirit of the application, and the scope of the present application is defined by the claims and their equivalents.

[0119] The above is a specific description of the preferred embodiments of the present application, but the present application is not limited to the described embodiments, and those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the present application, and these equivalent modifications or replacements are all included in the scope defined by the claims of the present application.

Claims

1. A vehicle trajectory image and trajectory coordinate processing method, characterized in that: include: Acquire a first trajectory image of the target vehicle, and perform preprocessing on the first trajectory image; The preprocessed first trajectory image is reconstructed using a variational autoencoder model to obtain a second trajectory image; performing quality assessment, color space conversion, and trajectory profile extraction on the second trajectory image in sequence to obtain a trajectory profile, and performing coordinate recognition and coordinate extraction on the trajectory profile to obtain coordinate points of the trajectory profile; The coordinate points are post-processed to obtain an original coordinate point set of the target vehicle, and a trajectory coordinate processing result is obtained based on the original coordinate point set.

2. The vehicle trajectory image and trajectory coordinate processing method according to claim 1, characterized in that: The preprocessing includes at least one of grayscale conversion, size normalization, and data enhancement.

3. The vehicle trajectory image and trajectory coordinate processing method according to claim 1, characterized in that: The training of the variational autoencoder model includes: The preprocessed dataset is used as input, and the model parameters are iteratively updated according to the loss function and Adam optimizer to perform model training and obtain the variational autoencoder model.

4. The vehicle trajectory image and trajectory coordinate processing method according to claim 3, characterized in that: The loss function includes a combination of reconstruction loss, KL divergence loss and color loss; The loss function is: in, is the loss function, is the reconstruction loss, is the KL divergence loss, is the color loss, α is the weight parameter for adjusting the KL divergence loss, and β is the parameter for adjusting the color loss; The reconstruction loss is used as the binary cross entropy between the second track image and the first track image; The KL divergence loss is the difference between the latent variable distribution and the prior distribution, and is expressed as: Among them, μ i represents the mean of the i-th latent variable, σ i represents the variance of the i-th latent variable; The color loss is a measure of the difference between the color channels in the second trajectory image and the color channels in the first trajectory image.

5. The vehicle trajectory image and trajectory coordinate processing method according to claim 4, characterized in that: The reconstructing the preprocessed first trajectory image using a variational autoencoder model to obtain a second trajectory image includes: The variational autoencoder model includes an encoder and a decoder, and the encoder maps the first trajectory image to a latent space; The latent vector is obtained by sampling in the latent space through the reparameterization method; The latent vector is reconstructed by the decoder to obtain a second trajectory image, wherein the encoder includes three linear layers, one of which flattens the input image and maps it to an intermediate dimensional space, and the other two linear layers respectively output the mean and logarithmic variance of the latent representation; the decoder includes two linear layers, and the decoder maps from the latent space back to the data space through the two linear layers and reconstructs the image.

6. The vehicle trajectory image and trajectory coordinate processing method according to claim 1, characterized in that: The step of sequentially performing quality assessment, color space conversion, and trajectory profile extraction on the second trajectory image to obtain a trajectory profile, and performing coordinate identification and coordinate extraction on the trajectory profile to obtain coordinate points of the trajectory profile includes: Performing a quality assessment on the second trajectory image using the MSE and PSNR indicator evaluation methods, wherein the quality assessment is used to determine the authenticity of the second trajectory image; Convert the second trajectory image from RGB color space to HSV color space; According to the preset color range, a mask is created in the HSV color space, and the trajectory contours of different colors are extracted according to the mask; The coordinate points of the trajectory contour are obtained by performing coordinate recognition and extraction through image processing algorithms.

7. The vehicle trajectory image and trajectory coordinate processing method according to claim 1, characterized in that: The post-processing of the coordinate points to obtain an original coordinate point set of the target vehicle, and obtaining a trajectory coordinate processing result based on the original coordinate point set, includes: The post-processing includes denoising and normalization, where the normalized x norm for: Where X is the extracted original coordinate point set, and x represents the coordinate point data in the original coordinate point set X; The original coordinate point set obtained by post-processing is saved as coordinate data to a preset format file to obtain the trajectory coordinate processing result.

8. A vehicle trajectory image and trajectory coordinate processing device, characterized in that: include: A first module is configured to obtain a first trajectory image of a target vehicle and perform preprocessing on the first trajectory image; The second module is used to reconstruct the preprocessed first trajectory image using a variational autoencoder model to obtain a second trajectory image; a third module, configured to sequentially perform quality assessment, color space conversion, and trajectory profile extraction on the second trajectory image to obtain a trajectory profile, and perform coordinate recognition and coordinate extraction on the trajectory profile to obtain coordinate points of the trajectory profile; The fourth module is used to perform post-processing on the coordinate points to obtain an original coordinate point set of the target vehicle, and obtain a trajectory coordinate processing result based on the original coordinate point set.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement the vehicle trajectory image and trajectory coordinate processing method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The storage medium stores a program, and the program is executed by a processor to implement the vehicle trajectory image and trajectory coordinate processing method according to any one of claims 1 to 7.

Images