WO2022151589A1 - Image enhancement method, apparatus and device, and storage medium - Google Patents

Abstract

An image enhancement method, apparatus and device, and a storage medium. The method comprises: a preset image recognition model recognizing on-site image data acquired from a traffic accident site, and determining whether the on-site image data meets a preset image standard; if the on-site image data does not meet the preset image standard, prompting a user to perform photographing again; if the on-site image data meets the preset image standard, conducting the next photographing task to obtain a target accident vehicle image of the traffic accident site; and inputting the target accident vehicle image into a preset image enhancement model for image enhancement, so as to generate a target enhanced image. According to the method, conversion from low-quality images to high-quality images can be completed merely by means of performing mapping function learning on high-quality pictures, the accuracy and the handling time effectiveness of vehicle insurance claim settlement are improved, and the technical problem of low claim settlement efficiency is solved.

Description

Image enhancement method, device, device and storage medium

This application claims the priority of the Chinese patent application with the application number 202110064519.6 and the invention title "Image Enhancement Method, Apparatus, Equipment and Storage Medium" filed with the China Patent Office on January 18, 2021, the entire contents of which are incorporated by reference in Applying.

technical field

The present application relates to the field of image processing, and in particular, to an image enhancement method, apparatus, device, and storage medium.

Background technique

With the popularization of personal mobile phones and the advancement of technology, the image processing process in the traditional auto insurance claims process usually requires a professional team, using more professional photographic equipment, and shooting and collecting image data for each area according to a complete collection process. . In the face of the increasing number of private cars and various small accidents, it has caused great pressure on auto insurance claims. Neither the number of personnel nor the processing efficiency can meet the current needs. Therefore, self-service services that have appeared in recent years , online submission of materials, quick claim settlement and even automatic claim settlement have become the iconic selling points of Internet insurance claim settlement services.

However, due to the quality of the images or video information collected in the step of shooting and uploading auto insurance claims materials, the accuracy of the subsequent steps will be directly determined, and ordinary users often do not have the operating skills of professionals, and the equipment used is ordinary mobile phones. The self-contained camera, and the shooting conditions of the location cannot be determined when car accident damage is determined, so the images or videos taken may often have problems such as wrong shooting target, unreasonable shooting angle, blurred image, insufficient exposure or strong light reflection, etc. People realize that these problems are much different from the images collected by professionals, which will increase the difficulty of OCR identification or manual review in the later stage, and cannot improve the processing efficiency.

SUMMARY OF THE INVENTION

The main purpose of this application is to improve the accuracy and timeliness of auto insurance claims settlement, and to solve the technical problem of low claim settlement efficiency.

In order to achieve the above purpose, a first aspect of the present application provides an image enhancement method, comprising: acquiring on-site image data of a traffic accident scene; identifying the on-site image data through a preset image recognition model, and judging the on-site image data Whether the preset image standard is met; if not, prompt the user to re-shoot; if so, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into the preset image enhancement The model performs image enhancement on the target accident vehicle image to generate a target enhanced image.

A second aspect of the present application provides an image enhancement device, comprising a memory, a processor, and computer-readable instructions stored on the memory and executable on the processor, the processor executing the computer-readable instructions When instructing, the following steps are implemented: acquiring on-site image data of the traffic accident scene; identifying the on-site image data through a preset image recognition model, and judging whether the on-site image data meets the preset image standard; if not, prompting the user to restart the Shooting; if yes, then enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into a preset image enhancement model, and perform image enhancement on the target accident vehicle image to generate Target-enhanced image.

A third aspect of the present application provides a computer-readable storage medium, where computer instructions are stored in the computer-readable storage medium, and when the computer instructions are executed on a computer, the computer is caused to perform the following steps: acquiring the traffic accident scene On-site image data; identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset image standards; if not, prompt the user to re-shoot; if so, enter the next shooting task , obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into a preset image enhancement model, perform image enhancement on the target accident vehicle image, and generate a target enhanced image.

A fourth aspect of the present application provides an image enhancement device, comprising: a first acquisition module for acquiring on-site image data of a traffic accident scene; a first judgment module for analyzing the on-site image data through a preset image recognition model Recognition is performed to determine whether the on-site image data meets the preset image standard; the prompting module is used to prompt the user to re-shoot when the on-site image data does not meet the preset image standard; the shooting module is used when the on-site image data does not meet the preset image standard. When the preset image standard is met, enter the next shooting task, and obtain the target accident vehicle image of the traffic accident scene; the image enhancement module is used for inputting the target accident vehicle image into a preset image enhancement model, and for the target accident vehicle image The vehicle image is image-enhanced to generate a target-enhanced image.

The present application uses a preset image recognition model to identify on-site image data obtained from the scene of a traffic accident to determine whether the on-site image data meets the preset image standards; if the on-site image data does not meet the preset image standards, the user is prompted to re-shoot; If the on-site image data meets the preset image standards, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into the preset image enhancement model for image enhancement to generate the target enhanced image. This solution only needs high-quality images to learn the mapping function to complete the conversion of low-quality to high-quality images, improve the accuracy and processing time of auto insurance claims, and solve the technical problem of low claim settlement efficiency.

Description of drawings

FIG. 1 is a schematic diagram of the first embodiment of the image enhancement method of the present application;

2 is a schematic diagram of a second embodiment of the image enhancement method of the present application;

3 is a schematic diagram of a third embodiment of the image enhancement method of the present application;

4 is a schematic diagram of a fourth embodiment of the image enhancement method of the present application;

5 is a schematic diagram of a fifth embodiment of the image enhancement method of the present application;

FIG. 6 is a schematic diagram of a first embodiment of an image enhancement apparatus of the present application;

FIG. 7 is a schematic diagram of a second embodiment of the image enhancement apparatus of the present application;

FIG. 8 is a schematic diagram of an embodiment of an image enhancement device of the present application.

Detailed ways

The embodiments of the present application provide an image enhancement method, device, device, and storage medium, which improve the accuracy and processing time of auto insurance claims, and solve the technical problem of low claim settlement efficiency.

For ease of understanding, the following describes the specific process of the embodiment of the present application, referring to FIG. 1 , the first embodiment of the image enhancement method in the embodiment of the present application includes:

101. Obtain on-site image data of a traffic accident scene;

In this embodiment, in order to ensure sufficient information is obtained, the user needs to take a 360-degree photograph around the accident vehicle and photograph key damaged parts at the scene of the traffic accident. For example, users need to take pictures of the corresponding parts of the vehicle on the spot through their personal mobile phones according to the guidelines, including the frame number, damaged parts, pictures of the whole vehicle with license plates, and pictures of the whole body of the vehicle.

102. Identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset image standard;

In this embodiment, due to the quality of the images or video information collected in the steps of shooting and uploading the auto insurance claim materials, the accuracy of the subsequent steps will be directly determined, and ordinary users often do not have the operating skills of professionals. The device is a camera that comes with an ordinary mobile phone, and the shooting conditions of the location cannot be determined when car insurance damage is determined. Therefore, the captured images or videos may often have wrong shooting targets, unreasonable shooting angles, blurred images, underexposure or strong light reflection. Therefore, it is necessary to identify the on-site image data of the accident vehicle taken by the user through the preset image recognition model, determine whether it is the corresponding photo required in the guide through the identification model, and quickly determine whether it conforms to the preset through solutions such as edge detection. standard.

103. When the on-site image data does not meet the preset image standard, prompt the user to re-shoot;

In this embodiment, it is determined locally whether the captured image conforms to the corresponding shooting standard, whether it is the corresponding photo required in the guide by the identification model, and whether there is blurring through solutions such as edge detection. If it does not meet the standard, the user will be prompted to shoot again.

104. When the on-site image data meets the preset image standard, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene;

In this embodiment, if the picture taken by the user on the spot meets the preset requirements, the next shooting task is entered to obtain a relatively high-quality image including brightness, contrast, white balance, latitude, noise, details, color transition, etc. , which is the target-enhanced image.

105. Input the target accident vehicle image into a preset image enhancement model, perform image enhancement on the target accident vehicle image, and generate a target enhanced image.

In this embodiment, the low-quality image of the target accident vehicle captured by the mobile phone is input into the target image enhancement model, and the comprehensive improvement including brightness, contrast, white balance, tolerance, noise, details, color transition, etc. is completed, and an image based on The original picture content and the enhanced picture that approximates the visual effect of professional digital cameras.

In this embodiment, the image enhancement method is to add some information or transform data to the original image by certain means, selectively highlight the interesting features in the image or suppress (mask) some unnecessary features in the image, so that the image is different from the original image. Visual response characteristics to match. In the process of image enhancement, the reason for image degradation is not analyzed, and the processed image is not necessarily close to the original image. Image enhancement technology can be divided into two categories: spatial domain-based algorithms and frequency domain-based algorithms according to the different spaces in which the enhancement process is performed.

In the embodiment of the present application, the on-site image data obtained from the traffic accident scene is identified by a preset image recognition model, and it is judged whether the on-site image data meets the preset image standard; if the on-site image data does not meet the preset image standard, the user is prompted Re-shoot; if the on-site image data meets the preset image standards, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into the preset image enhancement model for image enhancement, and generate the target enhanced image . This solution only needs high-quality images to learn the mapping function to complete the conversion of low-quality to high-quality images, improve the accuracy and processing time of auto insurance claims, and solve the technical problem of low claim settlement efficiency.

Referring to FIG. 2, the second embodiment of the image enhancement method in the embodiment of the present application includes:

201. Obtain historical vehicle images at the accident scene from a preset database;

In this embodiment, after a traffic accident occurs, a target image of at least one accident vehicle at the scene of the traffic accident is obtained, and these target images can reflect the specific situation of the accident scene, such as the license plate number, damaged part and degree of damage of the accident vehicle.

202. Use historical vehicle images as training samples to construct a training sample data set;

In this embodiment, a plurality of original image samples in various auto insurance claim scenarios can be obtained from the database first. The original image samples can be images taken by the user from the accident scene, or taken by a company professional with a professional camera images, etc. For example, photos taken by SLR cameras have less noise and richer details. Next, in response to a user operation, image enhancement is performed on the image parameters of each of the original image samples to generate an enhanced image sample after image enhancement. For example, each original image sample can be adjusted by a professional special effect designer to obtain an enhanced image sample after image enhancement, so that the enhanced image sample has better brightness and saturation.

203. Input the training sample data set into a preset image enhancement generator, and train the image enhancement generator to obtain a target image enhancement model;

In this embodiment, image enhancement refers to enhancing the useful information in the image, which may be a distortion process, the purpose of which is to improve the visual effect of the image, and is aimed at the application occasion of a given image. Purposefully emphasize the overall or local characteristics of the image, make the original unclear image clear or emphasize some interesting features, expand the difference between the features of different objects in the image, suppress the uninteresting features, and improve the image. Quality, rich information, strengthen image interpretation and recognition effect, to meet the needs of some special analysis.

In this embodiment, the training sample data is input into a preset image enhancement generator to train the image enhancement generator. The specific implementation steps include: inputting an original low-quality image x, and generating a The texture and color discriminator calculates the texture and color loss functions by comparing y with the image z captured by a professional camera, and trains the generator with secondary feedback. At the same time, the low-quality image y' is generated in reverse by applying the generator G' to the image y, and the generation control of the generator G is achieved by calculating the content loss with the original image x.

204. Acquire on-site image data of the traffic accident scene; 205. Identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset image standard; 206. When the on-site image data does not meet the preset image standard, Prompt the user to re-shoot; 207. When the on-site image data meets the preset image standard, enter the next shooting task to obtain the target accident vehicle image at the scene of the traffic accident; 208. Input the target accident vehicle image into the preset image enhancement model. The image of the accident vehicle is image-enhanced to generate a target-enhanced image.

Steps 204-208 in this embodiment are similar to steps 101-105 in the first embodiment, and are not repeated here.

In the embodiment of the present application, only high-quality pictures are needed to learn the mapping function to complete the conversion of low-quality to high-quality images, which improves the accuracy and processing time of auto insurance claims, and solves the technical problem of low claim settlement efficiency.

Referring to FIG. 3, the third embodiment of the image enhancement method in the embodiment of the present application includes:

301. Acquire historical vehicle images at the accident site from a preset database; 302. Use the historical vehicle images as training samples to construct a training sample data set; 303. Input the training sample data set into a preset image enhancement generator, and apply the image enhancement generator to the image enhancement generator. Carry out training to obtain a target image enhancement model; 304, obtain on-site image data of the traffic accident scene; 305, obtain a plurality of training pictures from a preset training picture database, and input the training pictures into the preset neural network model to obtain the training picture. predict class labels;

In this embodiment, the pictures of the accident vehicles taken at the scene of the vehicle accident in the past are obtained from the preset database as the pictures to be identified, and the predicted category labels of the pictures to be identified are obtained; wherein the pre-trained picture identification model adopts the convolutional neural network model , and the convolutional neural network model includes at least one layer of kernel pooling layer that upgrades the channel dimension from one-dimensional to multi-dimensional.

In this embodiment, in the process of training the image recognition model, the parameters in the image recognition model are constantly changing, that is, the image recognition model is constantly changing. In each training, the image recognition model used is the image recognition model updated in the previous training. However, when the image recognition model of this embodiment is used for the first time, each parameter adopts a preset initial value. The dimension of the predicted category label is determined during training. For example, if several training pictures in the training picture database include n pictures in total, the dimension of the predicted category label output by the picture recognition model can be set to n.

306. Generate a true category label of the training picture according to the category of the training picture;

In this embodiment, since the category of the training picture is known, the true category label of the category of the training picture can be generated according to the category of the training picture. Similarly, the true category label of the training picture is also in the form of a vector. In the vector, only the value of the element at the position corresponding to the category of the training picture is 1, and the rest of the positions are 0.

307. Generate a first loss function of the training picture according to the predicted category label of the training picture and the real category label of the training picture;

In this embodiment, according to the obtained predicted category label of the training picture and the obtained real category label of the training picture, the difference between the predicted category label of the training picture and the actual category label of the training picture can be known, and the training picture can be generated according to the difference. loss function. In the process of training the image recognition model in this embodiment, the value of the loss function should be gradually reduced, so that the trained image recognition model is more accurate. There are two or more loss functions in this embodiment, the feature difference values of the labels are calculated respectively, the multiple feature difference values calculated according to the corresponding loss functions are sorted, and the loss function corresponding to the feature difference value with the largest ranking is used as the final The loss function, which adjusts the weights back to the model.

In this embodiment, because the loss function represents a value descending in the gradient direction, the larger the loss function value, the less accurate the predicted category label output by the image recognition model after processing the input image. By continuously adjusting the loss function, the gradient of the final point of the loss function is 0. The smaller the value of the loss function, the more accurate the processing result of the image recognition model. That is, the derivative of the loss function of the training picture to each parameter in the picture recognition model is calculated, and each parameter in the picture recognition model is updated according to each parameter in the picture recognition model and the derivative of each parameter of the loss function corresponding to the training picture, so that Determine the image recognition model after this training.

The loss function, also known as the cost function, is a function that maps the value of a random event or its related random variables to non-negative real numbers to represent the "risk" or "loss" of the random event. In applications, loss functions are often associated with optimization problems as learning criteria, i.e. solving and evaluating models by minimizing the loss function. For example, it is used for parametric estimation of models in statistics and machine learning. In this embodiment, the loss function is pre-built and used to calculate the feature difference between the real feature and the predicted feature.

308. Update parameters in the neural network model according to the first loss function to obtain a target image recognition model;

In this embodiment, each parameter in the image recognition model is updated according to each parameter in the image recognition model and the derivative of the loss function of the training image to each parameter to obtain the target image recognition model.

Specifically, since the image recognition model includes multiple parameters, when the input training image is determined, the loss function can also be regarded as a function of the parameters in the image recognition model. At this time, the loss function of the training image can be calculated. The derivative of each parameter in the image recognition model; for each parameter in the image recognition model, when updating, the parameter in the image recognition model can be updated to the parameter minus the derivative of the loss function to the parameter, that is, every training Update the image recognition model.

Because the loss function represents a value that declines in the gradient direction, the larger the loss function value, the less accurate the predicted category label output by the image recognition model after processing the input image. In this embodiment, the loss is adjusted continuously. function so that the gradient of the final point of the loss function is 0. The smaller the value of the loss function, the more accurate the processing result of the image recognition model. In the technical solutions of the above embodiments, when the image recognition model is updated, it is specifically updated in a direction in which the loss function gradually decreases. Therefore, through continuous training, the loss function is continuously optimized, and the parameters of the image recognition model are continuously updated, so as to continuously update the image recognition model, so that the final image recognition model can recognize images very accurately.

309. Identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset image standard; 310. When the on-site image data does not meet the preset image standard, prompt the user to re-shoot; 311. When the on-site image data does not meet the preset image standard When the preset image standard is met, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; 312. Input the target accident vehicle image into the preset image enhancement model, perform image enhancement on the target accident vehicle image, and generate the target enhanced image .

Steps 301 - 303 and 309 - 312 in this embodiment are similar to steps 201 - 203 and 102 - 105 in the first embodiment, and are not repeated here.

In the embodiment of the present application, only high-quality pictures are needed to learn the mapping function to complete the conversion of low-quality to high-quality images, which improves the accuracy and processing time of auto insurance claims, and solves the technical problem of low claim settlement efficiency.

Referring to FIG. 4 , the fourth embodiment of the image enhancement method in the embodiment of the present application includes:

401. Acquire historical vehicle images at the accident site from a preset database; 402. Use the historical vehicle images as training samples to construct a training sample data set; 403. Perform image enhancement on the training sample data set to obtain a preset number of target original images and the corresponding enhanced image;

In this embodiment, a plurality of low-quality images are obtained from a preset database as training sample data, and these images are enhanced to obtain a preset number of original image samples and a corresponding preset number of enhanced image samples. Next, for each selected original image sample and a corresponding preset number of enhanced image samples, randomly crop an image of the target size at the same position in the original image sample and the corresponding enhanced image sample to obtain a preset number of The original image and the corresponding enhanced image of the target size.

404. Input the target original image into the CNN image enhancement network in the preset image enhancement generator to obtain an output image corresponding to the target original image;

In this embodiment, the collected target original images are input into the CNN image enhancement network in the preset image enhancement generator, and an output image corresponding to each target original image can be obtained.

405. Determine a second loss function between the output image and the enhanced image;

In this embodiment, the loss function is pre-built and is used to calculate the feature difference between the real feature and the predicted feature, and sorts the multiple feature differences calculated according to the corresponding loss function, and sorts the feature difference with the largest one. The loss function corresponding to the value is used as the final loss function. After the loss function value is obtained, the training of the CNN image enhancement network is supervised and the network parameters are updated based on the loss function value.

In this embodiment, the output image and the enhanced image may be blurred by using the mean blurring method, and the mean error value of the blurred output image and the enhanced image may be calculated to obtain the color_loss value. By using the mean blurring method to blur the output image and the enhanced image, the interference of high-frequency information is eliminated, and the network learns more color information. In actual implementation, the mean pooling layer in the CNN image enhancement network can be used and stride is set to 1 to realize end2end training, so as to perform l2loss (ie Mean Square Error Loss) on the blurred output image and the enhanced image to obtain Get the color_loss value.

Further, the error calculation is performed on the output image and the enhanced image through the feature map output by the CNN image enhancement network in the same layer to obtain the vgg_loss value. Among them, vgg_loss is a semantic type of loss, which can better generate semantic information. Exemplarily, this embodiment may use vgg19 as a network structure for generating feature maps of different layers, and at the same time initialize the vgg19 network with network parameters trained on the ImageNet dataset. The output image and the enhanced image are subjected to l2loss on the feature map output by the same layer through the vgg19 network to obtain the vgg_loss value.

Further, the grayscale image of the output image and the grayscale image of the enhanced image are obtained respectively, and an error calculation is performed on the grayscale image of the output image and the grayscale image of the enhanced image to obtain a texture_loss value. texture_loss, also known as texture loss, can add texture details on the basis of the above to ensure image enhancement without losing detail information.

406. Based on the second loss function, iteratively train the CNN image enhancement network in the image enhancement generator until the CNN image enhancement network converges to obtain the target image enhancement model;

In this embodiment, after the loss function value is obtained, the training of the CNN image enhancement network is supervised and network parameters are updated based on the loss function value, until the CNN image enhancement network converges to obtain the target image enhancement model;

In this embodiment, the method of judging whether the CNN image enhancement network reaches the training convergence condition may be: judging whether the change value of the loss value is less than a preset value, and if the change value of the loss value is less than the preset value, then It is determined that the CNN image enhancement network has reached the training convergence condition, otherwise it is determined that the CNN image enhancement network has not reached the training convergence condition. The preset value can be set according to actual needs. For example, the preset value can be set to a value close to 0, or can also be set to 0. If the preset value is 0, the loss value does not change. When it is determined that the CNN image enhancement network has reached the training convergence condition, otherwise it is determined that the CNN image enhancement network has not reached the training convergence condition. If the CNN image enhancement network does not reach the training convergence condition, repeat the above steps to continue training.

When the CNN image enhancement network reaches the training convergence condition, the network parameters of the CNN image enhancement network are updated, thereby outputting the corresponding image enhancement model. The image enhancement model can perform image enhancement on the input image. In detail, after receiving the to-be-processed image, the to-be-processed image may be input into the image enhancement model to obtain an enhanced image corresponding to the to-be-processed image. For example, the user inputs the image captured by the mobile phone as the image to be processed into the trained image enhancement model, and then obtains the corresponding enhanced image through the layer-by-layer calculation of the network, and the corresponding enhanced image is the target enhanced image.

407. Acquire on-site image data of the traffic accident scene; 408. Identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset image standard; 409. When the on-site image data does not meet the preset image standard, Prompt the user to re-shoot; 410. When the on-site image data meets the preset image standard, enter the next shooting task, and obtain the target accident vehicle image at the traffic accident scene; 411. Input the target accident vehicle image into the preset image enhancement model, and the target accident vehicle image The image of the accident vehicle is image-enhanced to generate a target-enhanced image.

Steps 407-411 in this embodiment are similar to steps 101-105 in the first embodiment, and are not repeated here.

In the embodiment of the present application, only high-quality pictures are needed to learn the mapping function to complete the conversion of low-quality to high-quality images, which improves the accuracy and processing time of auto insurance claims, and solves the technical problem of low claim settlement efficiency.

Referring to FIG. 5 , the fifth embodiment of the image enhancement method in the embodiment of the present application includes:

501. Acquire on-site image data of the traffic accident scene; 502. Identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset image standard; 503. When the on-site image data does not meet the preset image standard, Prompt the user to re-shoot; 504. When the on-site image data meets the preset image standard, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; 505. Perform down-sampling processing on the target accident vehicle image to obtain a down-sampled image ;

In this embodiment, when image enhancement processing needs to be performed on a certain image, the image can be used as the target enhancement image to be enhanced. The target-enhanced image may be a high-resolution image. Moreover, the target-enhanced image can be a single-channel image or a multi-channel image. Wherein, the color space of the multi-channel image may be: RGB (Red, Green, Blue, red, green, blue), YUV (Luminance, Chrominance, Chroma, brightness, chromaticity, density) or other color spaces. Not limited. The electronic device can take an enhanced image of the target through a built-in or external camera, and can also communicate with other devices to receive enhanced images of the target sent by other devices. The present disclosure does not limit the manner in which the electronic device obtains the target enhanced image.

After the electronic device acquires the target enhanced image, it can perform down-sampling processing on the target enhanced image, thereby obtaining the down-sampled image. The obtained down-sampled image can be further input into a deep learning network, and the deep learning network performs image enhancement processing on the down-sampled image.

In this embodiment, the target enhanced image may be down-sampled by x times, so that the resolution, width and height of the obtained down-sampled image are all 1/x times the corresponding parameters of the target enhanced image.

506. Input the down-sampled image into the target image enhancement model to obtain image enhancement data corresponding to the target image enhancement model;

In this embodiment, in order to perform image enhancement processing on the down-sampled image, the deep learning network can be trained according to the sample image and the sample enhanced image corresponding to the sample image in advance to obtain a trained deep learning network. Among them, the sample images and the corresponding sample-enhanced images are training samples used for training. Sample images and corresponding sample-enhanced images can be obtained from the existing training library as training samples. The sample-enhanced image may be obtained by using a multi-exposure fusion method, or may be obtained by using other single-frame image enhancement methods, which is not limited in the present disclosure.

In this embodiment, the structure of the deep learning network may be any existing deep learning network model. Specifically, during training, the number of training samples can be determined according to actual needs. At the same time, a reasonable loss function or objective function and corresponding target value can also be set to determine whether the deep learning network is well trained.

After the deep learning network is trained, the network parameters of the deep learning network are determined. Since the deep learning network is trained according to the sample image and the corresponding sample enhanced image, after the down-sampled image is input into the trained deep learning network, the image enhancement data corresponding to the down-sampled image can be obtained. The image enhancement data is data representing the degree of enhancement of the downsampled image enhanced image relative to the downsampled image. The image enhancement data can be in various forms. For example, the image enhancement data may include mapping parameters for mapping each pixel in the down-sampled image to a corresponding pixel, wherein the corresponding pixel of any pixel is the down-sampled image. The pixel in the image after image enhancement has the same position as the pixel. As another example, the image enhancement data may include a downsampled image enhanced image. Moreover, it can be understood that the output result of the deep learning network is of the same type as the image enhancement data corresponding to the downsampled image, that is, when the output result of the deep learning network is the mapping parameter, the image enhancement data corresponding to the downsampled image is of the same type. is the mapping parameter, and when the output result of the deep learning network is an enhanced image, the image enhancement data corresponding to the downsampled image is the image enhanced by the downsampled image.

507. Determine the matching point of each pixel in the target accident vehicle image in the down-sampled image;

In this embodiment, since the resolution of the down-sampled image is lower than the resolution of the target enhanced image, the image enhancement data corresponding to the down-sampled image does not correspond to the target enhanced image. In order to obtain the enhanced image corresponding to the target enhanced image, in this embodiment, the matching point of each pixel in the target enhanced image in the down-sampled image may be determined first, then the target enhancement parameter corresponding to the pixel is determined, and finally , and use the target enhancement parameter to adjust the pixel value of the pixel.

In this embodiment, for each pixel in the target enhanced image, the corresponding pixel in the down-sampled image is determined, and in the search area with the corresponding pixel as the center and the size is M×N, find the corresponding pixel in the down-sampled image. For the pixel point with the smallest absolute value of the difference between the pixel values of the pixel point, the found pixel point is used as the matching point of the pixel point in the down-sampled image.

508. Determine the target enhancement parameter corresponding to the pixel point based on the enhancement data corresponding to the matching point of the pixel point in the image enhancement data;

In this embodiment, after determining the matching point of the pixel in the down-sampled image for each pixel in the target enhancement image, the target enhancement parameter corresponding to the pixel can be determined based on the enhancement data corresponding to the matching point , and further adjust the pixel value of the pixel based on the target enhancement parameter. Specifically, the form of the image enhancement data is different, and the way of determining the target enhancement parameter corresponding to each pixel in the target enhancement image is also different. Two ways are given below for description.

Optionally, the image enhancement data may include: each pixel in the down-sampled image is mapped to a mapping parameter of a corresponding pixel, and the corresponding pixel of any pixel is the same as the pixel in the enhanced image of the down-sampled image. of pixels. The so-called mapping a pixel point to a corresponding pixel point specifically refers to: adjusting the pixel value of a pixel point to the pixel value of the corresponding pixel point.

Correspondingly, for each pixel point in the target enhanced image, determining the target enhancement parameter corresponding to the pixel point based on the enhancement data corresponding to the matching point of the pixel point in the image enhancement data, may include: For each pixel, the target parameter corresponding to the matching point of the pixel is determined from the respective mapping parameters, and the determined target parameter is used as the target enhancement parameter corresponding to the pixel.

509. Based on the target enhancement parameter corresponding to the pixel point, adjust the pixel value of the pixel point to obtain the target enhanced image corresponding to the target accident vehicle image;

In this embodiment, for each pixel in the target enhancement image, after the target enhancement parameter corresponding to the pixel is determined, the pixel value of the pixel can be adjusted based on the target enhancement parameter. This adjustment process is the process of image enhancement processing on the target enhanced image.

510. Acquire vehicle owner identity information of the user and damage information of the accident vehicle, where the vehicle information includes auto insurance information of the accident vehicle;

In this embodiment, the accident type, the relative position information of the accident vehicle, the first vehicle identifier of the accident vehicle, the damaged position of the accident vehicle, and the damage degree of the accident vehicle are sent to the described client.

The accident type is the premise for determining the damage of the accident vehicle. Different accident types will affect the damaged position and damage degree of the accident vehicle. For rear-end collisions, the damaged position of the accident vehicle is the front and rear ends of the accident vehicle, and most of them are collision damage, and the degree of damage is relatively heavier than the scratch accident; for parallel collisions and scratch accidents, the damaged position of the accident vehicle It is the two sides of the accident vehicle, and most of them are scratch damage; for a cornering collision accident, the damaged position of the accident vehicle is one side and the front end of the accident vehicle, and there are both collision damage and scratch damage. Therefore, it is necessary to determine the accident type through the accident image. According to the owner's identity information entered by the user and the damage information of the accident vehicle, the vehicle is claimed.

511. Receive the claim settlement request uploaded by the user, and input the target enhanced image into the trained convolutional neural network model for prediction, and obtain the claim settlement probability of the target enhanced image;

In this embodiment, the claim settlement request uploaded by the user includes the target enhanced image (that is, the target enhanced image in this embodiment) and the claim information, wherein the target enhanced image is the user taking pictures of the damaged part of the vehicle when a vehicle accident occurs For example, a picture of a rear-end collision of a vehicle, a picture of a rollover of a vehicle, and a picture of a vehicle scratching. Claim information refers to the user uploading information related to claims to the insurance company after a vehicle accident, such as the car model, the cause of damage, and the policy number, etc. After a vehicle accident occurs, the user fills in the basic vehicle information and the cause of the accident through the mobile terminal 10, takes a picture of the damaged vehicle on site, and uploads the target enhanced image and claim information to the insurance company's server to generate a claim request. After receiving the claim settlement request uploaded by the user, the trained convolutional neural network model is called, and the target enhanced image is input into the convolutional neural network model for prediction, and the claim settlement probability of the target enhanced image is obtained, that is, the target The probability of whether the enhanced image should be claimed.

512. Determine whether the claim settlement probability of the target enhanced image is greater than a preset threshold;

In this embodiment, a classifier SVM (Support Vector Machine) is used for classification, which is a two-class model for classifying the target enhanced image into two classes, one class is classified as belonging to claim settlement, and the other class is classified as not belonging to claim settlement. When the one-dimensional feature vector is obtained, the one-dimensional feature vector is input into the classifier, and the classifier maps the one-dimensional feature vector to a value in the range of 0 to 1, which is the claim settlement probability.

513. When the claim settlement probability of the target enhanced image is greater than a preset threshold, determine that the target enhanced image can be claimed;

In this embodiment, after the claim settlement probability of the target enhanced image is obtained, the claim settlement probability is compared with a preset threshold, and the preset threshold is 0.7. Of course, it can be understood that it can be any other value. When the claim settlement probability corresponding to the target enhanced image is greater than the preset threshold, it means that the target enhanced image is very close to the target accident vehicle image that has already been claimed, and it is determined that the target enhanced image is claimable. When the claim settlement probability corresponding to the target enhanced image is less than the preset threshold, it indicates that the target enhanced image has a risk of insurance fraud or fraud, and it is determined that the target enhanced image is not eligible for claim settlement. For example, when the claim settlement probability is 0.8, then the claim settlement probability is greater than the preset threshold, and it is determined that the target enhanced image can be claimed.

514. Based on the user's owner identity information and the identification information of the accident vehicle, make a claim for the accident vehicle.

In this embodiment, all claim settlement cases are screened from the preset database, claim settlement cases that have already been settled are selected, and claim settlement information of the screened claim settlement cases is obtained. Claims that have been settled can be obtained by selecting the marked claim cases; then, the claim information of the screened claim cases and the claim information uploaded by the user are matched according to preset rules, wherein the preset rules are, for example, according to the vehicle. Match the model, vehicle age, and damaged parts. For example, if the vehicle model uploaded by the user is type A, the vehicle age is 2 years, and the damaged part is the rear of the vehicle, the corresponding claim information will be searched for one-to-one matching. Claim settlement information, and use the claimed amount in the matching claim settlement information as the estimated claim settlement amount, and send the estimated claim settlement amount to the user.

Steps 501-504 in this embodiment are similar to steps 101-104 in the first embodiment, and are not repeated here.

In the embodiment of the present application, the on-site image data obtained from the traffic accident scene is identified by a preset image recognition model, and it is judged whether the on-site image data meets the preset image standard; if the on-site image data does not meet the preset image standard, a prompt is displayed The user re-shoots; if the on-site image data meets the preset image standards, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into the preset image enhancement model for image enhancement to generate target enhancement image. This solution only needs high-quality images to learn the mapping function to complete the conversion of low-quality to high-quality images, improve the accuracy and processing time of auto insurance claims, and solve the technical problem of low claim settlement efficiency.

The image enhancement method in the embodiment of the present application has been described above, and the image enhancement device in the embodiment of the present application is described below. Please refer to FIG. 6 . The first embodiment of the image enhancement device in the embodiment of the present application includes: a first acquisition module 601, for acquiring on-site image data of a traffic accident scene; a first judgment module 602, for identifying the on-site image data through a preset image recognition model, and judging whether the on-site image data satisfies the preset image standard; prompting The module 603 is used for prompting the user to re-shoot when the on-site image data does not meet the preset image standard; the shooting module 604 is used for entering the next shooting task when the on-site image data meets the preset image standard, and obtaining the The target accident vehicle image at the scene of the traffic accident; the image enhancement module 605 is configured to input the target accident vehicle image into a preset image enhancement model, perform image enhancement on the target accident vehicle image, and generate a target enhanced image.

In the embodiment of the present application, only high-quality pictures are needed to learn the mapping function to complete the conversion of low-quality to high-quality images, which improves the accuracy and processing time of auto insurance claims, and solves the technical problem of low claim settlement efficiency.

Please refer to FIG. 7 , the second embodiment of the image enhancement apparatus in the embodiment of the present application, the image enhancement apparatus specifically includes:

The first acquisition module 601 is used to acquire the on-site image data of the traffic accident scene; the first judgment module 602 is used to identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset requirements. Image standard; the prompting module 603 is used to prompt the user to re-shoot when the on-site image data does not meet the preset image standard; the shooting module 604 is used to enter the next shooting when the on-site image data meets the preset image standard The task is to obtain the target accident vehicle image at the traffic accident scene; the image enhancement module 605 is used to input the target accident vehicle image into a preset image enhancement model, and perform image enhancement on the target accident vehicle image to generate a target enhanced image. .

In this embodiment, the image enhancement device further includes:

The second acquisition module 605 is used to acquire historical vehicle images of the accident scene from the preset database; the construction module 606 is used to construct a training sample data set using the historical vehicle images as training samples; the training module 607 is used to The training sample data set is input to a preset image enhancement generator, and the image enhancement generator is trained to obtain a target image enhancement model.

In this embodiment, the image enhancement device further includes:

The input module 608 is used to obtain a plurality of training pictures from the preset training picture database, and the training pictures are preset to the neural network model to obtain the predicted category label of the training pictures; the generation module 609 is used for according to the The category of the training picture, the true category label of the training picture is generated; the first loss function of the training picture is generated according to the predicted category label of the training picture and the true category label of the training picture; the update module 610 uses The parameters in the neural network model are updated according to the first loss function to obtain a target image recognition model.

In this embodiment, the training module 607 is specifically used for:

The image enhancement unit 6071 is used to perform image enhancement on the training sample data set to obtain a preset number of target original images and corresponding enhanced images; the input unit 6072 is used to input the target original image into the preset image enhancement In the CNN image enhancement network in the generator, the output image corresponding to the target original image is obtained; the determining unit 6073 is used to determine the second loss function between the output image and the enhanced image; the iterative training unit 6074 is used for Based on the second loss function, the CNN image enhancement network in the image enhancement generator is performed until the CNN image enhancement network converges to obtain a target image enhancement model.

In this embodiment, the image enhancement module 605 is specifically used for:

Perform down-sampling processing on the target accident vehicle image to obtain a down-sampled image; input the down-sampled image into the target image enhancement model to obtain image enhancement data corresponding to the target accident vehicle image; determine the target accident vehicle a matching point of each pixel in the image in the down-sampled image; based on the enhancement data corresponding to the matching point of the pixel in the image enhancement data, determine the target enhancement parameter corresponding to the pixel; based on The target enhancement parameter corresponding to the pixel point is adjusted, and the pixel value of the pixel point is adjusted to obtain the target enhancement image corresponding to the target accident vehicle image.

In this embodiment, the image enhancement device further includes:

The third obtaining module 611 is configured to obtain the vehicle owner identity information of the user and the damage information of the accident vehicle, wherein the vehicle information includes the car insurance information of the accident vehicle; the prediction module 612 is configured to receive the upload from the user and input the target enhanced image into the trained convolutional neural network model for prediction, and obtain the claim settlement probability of the target enhanced image; the second judgment module 613 is used to judge the claim settlement of the target enhanced image Whether the probability is greater than a preset threshold; the determination module 614 is used to determine that the target enhanced image can be claimed when the claim settlement probability of the target enhanced image is greater than the preset threshold; the claim settlement module 615 is used based on the user's vehicle owner identity information and the identification information of the accident vehicle, and make a claim for the accident vehicle.

In the embodiment of the present application, a preset image recognition model is used to identify the on-site image data obtained from the scene of a traffic accident to determine whether the on-site image data meets the preset image standard; if the on-site image data does not meet the preset image standard, the user is prompted to Re-shoot; if the on-site image data meets the preset image standards, enter the next shooting task to obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into the preset image enhancement model for image enhancement, and generate the target enhanced image . This solution only needs high-quality pictures to learn the mapping function to complete the conversion of low-quality to high-quality images, improve the accuracy and processing time of auto insurance claims, and solve the technical problem of low claim settlement efficiency.

Figures 6 and 7 above describe in detail the image enhancement apparatus in the embodiment of the present application from the perspective of modular functional entities, and the following describes the image enhancement device in the embodiment of the present application in detail from the perspective of hardware processing.

8 is a schematic structural diagram of an image enhancement device provided by an embodiment of the present application. The image enhancement device 800 may vary greatly due to different configurations or performance, and may include one or more processors (central processing units, CPUs) ) 810 (eg, one or more processors) and memory 820, one or more storage media 830 (eg, one or more mass storage devices) storing application programs 833 or data 832. Among them, the memory 820 and the storage medium 830 may be short-term storage or persistent storage. The program stored in the storage medium 830 may include one or more modules (not shown in the figure), and each module may include a series of instructions to operate on the image enhancement apparatus 800 . Further, the processor 810 may be configured to communicate with the storage medium 830, and execute a series of instruction operations in the storage medium 830 on the image enhancement device 800, so as to implement the steps of the image enhancement method provided by the above method embodiments.

Image enhancement device 800 may also include one or more power supplies 840, one or more wired or wireless network interfaces 850, one or more input and output interfaces 860, and/or, one or more operating systems 831, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD and more. Those skilled in the art can understand that the structure of the image enhancement device shown in FIG. 8 does not constitute a limitation to the image enhancement device provided in this application, and may include more or less components than those shown in the figure, or combine some components, or different component layout.

The present application also provides an image enhancement device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, the memory and the at least one processor are interconnected by a line; the at least one processor calls the at least one processor The instructions in the memory are used to cause the image enhancement device to perform the steps in the image enhancement method described above.

The present application also provides a computer-readable storage medium, and the computer-readable storage medium may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on the computer, the computer performs the following steps:

Obtain on-site image data of the traffic accident scene; identify the on-site image data through a preset image recognition model, and determine whether the on-site image data meets the preset image standard; if not, prompt the user to re-shoot; if so, enter the The next shooting task is to obtain the target accident vehicle image at the traffic accident scene; input the target accident vehicle image into a preset image enhancement model, and perform image enhancement on the target accident vehicle image to generate a target enhanced image.

The technical solution of the present application can be embodied in the form of a software product in essence, or the part that contributes to the prior art, or all or part of the technical solution. The computer software product is stored in a storage medium, including a number of instructions for So that a computer device (which may be a personal computer, a server, or a network device, etc.) executes all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, removable hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk and other media that can store program codes .

As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand: The technical solutions described in the embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions in the embodiments of the present application.

Claims (20)

1. An image enhancement method comprising:

   Obtain the scene image data of the traffic accident scene;

   Identifying the on-site image data through a preset image recognition model, and judging whether the on-site image data satisfies the preset image standard;

   If not, prompt the user to shoot again;

   If so, enter the next photographing task to obtain the target accident vehicle image at the traffic accident scene;

   The target accident vehicle image is input into a preset image enhancement model, and image enhancement is performed on the target accident vehicle image to generate a target enhanced image.
2. The image enhancement method according to claim 1, wherein before acquiring the scene image data of the traffic accident scene, the method further comprises:

   Obtain historical vehicle images at the accident scene from a pre-built database;

   Using the historical vehicle images as training samples to construct a training sample data set;

   The training sample data set is input into a preset image enhancement generator, and the image enhancement generator is trained to obtain a target image enhancement model.
3. The image enhancement method according to claim 1, wherein before identifying the on-site image data by using a preset image recognition model and judging whether the on-site image data satisfies a preset image standard, the method further comprises:

   Obtain a plurality of training pictures from the preset training picture database, and input the training pictures into the preset neural network model to obtain the predicted category label of the training pictures;

   According to the category of the training picture, generate the true category label of the training picture;

   Generate the first loss function of the training picture according to the predicted category label of the training picture and the real category label of the training picture;

   The parameters in the neural network model are updated according to the first loss function to obtain a target image recognition model.
4. The image enhancement method according to claim 2, wherein the inputting the training sample data set into a preset image enhancement generator, and training the image enhancement generator to obtain the target image enhancement model comprises:

   Perform image enhancement on the training sample data set to obtain a preset number of target original images and corresponding enhanced images;

   Inputting the target original image into the CNN image enhancement network in the preset image enhancement generator to obtain an output image corresponding to the target original image;

   determining a second loss function between the output image and the enhanced image;

   Based on the second loss function, the CNN image enhancement network in the image enhancement generator is iteratively trained until the CNN image enhancement network converges to obtain a target image enhancement model.
5. The image enhancement method according to claim 1, wherein the inputting the target accident vehicle image into a preset image enhancement model, performing image enhancement on the target accident vehicle image, and generating the target enhanced image comprises:

   Perform down-sampling processing on the target accident vehicle image to obtain a down-sampled image;

   Inputting the down-sampled image into the target image enhancement model to obtain image enhancement data corresponding to the target accident vehicle image;

   determining the matching point of each pixel in the target accident vehicle image in the down-sampled image;

   Determine the target enhancement parameter corresponding to the pixel point based on the enhancement data corresponding to the matching point of the pixel point in the image enhancement data;

   Based on the target enhancement parameter corresponding to the pixel point, the pixel value of the pixel point is adjusted to obtain the target enhancement image corresponding to the target accident vehicle image.
6. The image enhancement method according to claim 1, wherein after inputting the target accident vehicle image into a preset image enhancement model, performing image enhancement on the target accident vehicle image, and generating the target enhanced image, the method further comprises:

   Acquiring vehicle owner identity information of the user and damage information of the accident vehicle, wherein the vehicle information includes auto insurance information of the accident vehicle;

   Receive the claim settlement request uploaded by the user, input the target enhanced image into the trained convolutional neural network model for prediction, and obtain the claim settlement probability of the target enhanced image;

   judging whether the claim settlement probability of the target enhanced image is greater than a preset threshold;

   If the claim settlement probability of the target enhanced image is greater than a preset threshold, determining that the target enhanced image can be claimed;

   A claim is made for the accident vehicle based on the vehicle owner identity information of the user and the identification information of the accident vehicle.
7. An image enhancement device, comprising a memory, a processor, and computer-readable instructions stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer-readable instructions:

   Obtain the scene image data of the traffic accident scene;

   Identifying the on-site image data through a preset image recognition model, and judging whether the on-site image data satisfies the preset image standard;

   If not, prompt the user to shoot again;

   If so, enter the next photographing task to obtain the target accident vehicle image at the traffic accident scene;

   The target accident vehicle image is input into a preset image enhancement model, and image enhancement is performed on the target accident vehicle image to generate a target enhanced image.
8. The image enhancement device according to claim 7, wherein the processor further implements the following steps when executing the computer program:

   Obtain historical vehicle images at the accident scene from a pre-built database;

   Using the historical vehicle images as training samples to construct a training sample data set;

   The training sample data set is input into a preset image enhancement generator, and the image enhancement generator is trained to obtain a target image enhancement model.
9. The image enhancement device according to claim 7, wherein the processor further implements the following steps when executing the computer program:

   Obtain a plurality of training pictures from the preset training picture database, and input the training pictures into the preset neural network model to obtain the predicted category label of the training pictures;

   According to the category of the training picture, generate the true category label of the training picture;

   Generate the first loss function of the training picture according to the predicted category label of the training picture and the real category label of the training picture;

   The parameters in the neural network model are updated according to the first loss function to obtain a target image recognition model.
10. The image enhancement device according to claim 8, wherein the processor further implements the following steps when executing the computer program:

    Perform image enhancement on the training sample data set to obtain a preset number of target original images and corresponding enhanced images;

    Inputting the target original image into the CNN image enhancement network in the preset image enhancement generator to obtain an output image corresponding to the target original image;

    determining a second loss function between the output image and the enhanced image;

    Based on the second loss function, the CNN image enhancement network in the image enhancement generator is iteratively trained until the CNN image enhancement network converges to obtain a target image enhancement model.
11. The image enhancement device according to claim 7, wherein the processor further implements the following steps when executing the computer program:

    Perform down-sampling processing on the target accident vehicle image to obtain a down-sampled image;

    Inputting the down-sampled image into the target image enhancement model to obtain image enhancement data corresponding to the target accident vehicle image;

    determining the matching point of each pixel in the target accident vehicle image in the down-sampled image;

    Determine the target enhancement parameter corresponding to the pixel point based on the enhancement data corresponding to the matching point of the pixel point in the image enhancement data;

    Based on the target enhancement parameter corresponding to the pixel point, the pixel value of the pixel point is adjusted to obtain the target enhancement image corresponding to the target accident vehicle image.
12. The image enhancement device according to claim 7, wherein the processor further implements the following steps when executing the computer program:

    Acquiring vehicle owner identity information of the user and damage information of the accident vehicle, wherein the vehicle information includes auto insurance information of the accident vehicle;

    Receive the claim settlement request uploaded by the user, input the target enhanced image into the trained convolutional neural network model for prediction, and obtain the claim settlement probability of the target enhanced image;

    judging whether the claim settlement probability of the target enhanced image is greater than a preset threshold;

    If the claim settlement probability of the target enhanced image is greater than a preset threshold, determining that the target enhanced image can be claimed;

    A claim is made for the accident vehicle based on the vehicle owner identity information of the user and the identification information of the accident vehicle.
13. A computer-readable storage medium, storing computer instructions in the computer-readable storage medium, when the computer instructions are executed on a computer, the computer is made to perform the following steps:

    Obtain the scene image data of the traffic accident scene;

    Identifying the on-site image data through a preset image recognition model, and judging whether the on-site image data satisfies the preset image standard;

    If not, prompt the user to shoot again;

    If so, enter the next photographing task to obtain the target accident vehicle image at the traffic accident scene;

    The target accident vehicle image is input into a preset image enhancement model, and image enhancement is performed on the target accident vehicle image to generate a target enhanced image.
14. The computer-readable storage medium of claim 13, when the computer instructions are executed on a computer, causing the computer to further perform the following steps:

    Obtain historical vehicle images at the accident scene from a pre-built database;

    Using the historical vehicle images as training samples to construct a training sample data set;

    The training sample data set is input into a preset image enhancement generator, and the image enhancement generator is trained to obtain a target image enhancement model.
15. The computer-readable storage medium of claim 13, when the computer instructions are executed on a computer, causing the computer to further perform the following steps:

    Obtain a plurality of training pictures from the preset training picture database, and input the training pictures into the preset neural network model to obtain the predicted category label of the training pictures;

    According to the category of the training picture, generate the true category label of the training picture;

    Generate the first loss function of the training picture according to the predicted category label of the training picture and the real category label of the training picture;

    The parameters in the neural network model are updated according to the first loss function to obtain a target image recognition model.
16. The computer-readable storage medium of claim 14, which, when executed on a computer, causes the computer to further perform the following steps:

    Perform image enhancement on the training sample data set to obtain a preset number of target original images and corresponding enhanced images;

    Inputting the target original image into the CNN image enhancement network in the preset image enhancement generator to obtain an output image corresponding to the target original image;

    determining a second loss function between the output image and the enhanced image;

    Based on the second loss function, the CNN image enhancement network in the image enhancement generator is iteratively trained until the CNN image enhancement network converges to obtain a target image enhancement model.
17. The computer-readable storage medium of claim 13, which, when executed on a computer, causes the computer to further perform the following steps:

    Perform down-sampling processing on the target accident vehicle image to obtain a down-sampled image;

    Inputting the down-sampled image into the target image enhancement model to obtain image enhancement data corresponding to the target accident vehicle image;

    determining the matching point of each pixel in the target accident vehicle image in the down-sampled image;

    Determine the target enhancement parameter corresponding to the pixel point based on the enhancement data corresponding to the matching point of the pixel point in the image enhancement data;

    Based on the target enhancement parameter corresponding to the pixel point, the pixel value of the pixel point is adjusted to obtain the target enhancement image corresponding to the target accident vehicle image.
18. The computer-readable storage medium of claim 13, when the computer instructions are executed on a computer, causing the computer to further perform the following steps:

    Acquiring vehicle owner identity information of the user and damage information of the accident vehicle, wherein the vehicle information includes auto insurance information of the accident vehicle;

    Receive the claim settlement request uploaded by the user, input the target enhanced image into the trained convolutional neural network model for prediction, and obtain the claim settlement probability of the target enhanced image;

    judging whether the claim settlement probability of the target enhanced image is greater than a preset threshold;

    If the claim settlement probability of the target enhanced image is greater than a preset threshold, determining that the target enhanced image can be claimed;

    A claim is made for the accident vehicle based on the vehicle owner identity information of the user and the identification information of the accident vehicle.
19. An image enhancement device, wherein the image enhancement device comprises:

    The first acquisition module is used to acquire the scene image data of the traffic accident scene;

    a first judging module, configured to identify the on-site image data through a preset image recognition model, and judge whether the on-site image data satisfies a preset image standard;

    a prompting module, used for prompting the user to re-shoot when the on-site image data does not meet the preset image standard;

    a photographing module, configured to enter the next photographing task when the on-site image data meets a preset image standard, and obtain an image of the target accident vehicle at the traffic accident scene;

    The image enhancement module is used for inputting the target accident vehicle image into a preset image enhancement model, and performing image enhancement on the target accident vehicle image to generate a target enhanced image.
20. The image enhancement device of claim 19, wherein the image enhancement device further comprises:

    The second acquisition module is used to acquire historical vehicle images of the accident scene from the preset database;

    a building module for constructing a training sample data set using the historical vehicle images as training samples;

    A training module, configured to input the training sample data set into a preset image enhancement generator, and train the image enhancement generator to obtain a target image enhancement model.

Images