Image processing method and device and readable storage medium

Abstract

The invention provides an image processing method and device and a readable storage medium. The method comprises the following steps: acquiring a to-be-processed image; performing image transformation on the to-be-processed image to obtain N first images; the size of the first image is the same as that of the to-be-processed image; n is an integer greater than or equal to 2; processing the N first images by using an auto-encoder model to obtain second images corresponding to the first images; the self-encoder model is used for carrying out deblurring processing on an image; performing image inverse transformation on each second image to obtain a third image corresponding to each second image; according to each third image, obtaining a processed image; the processed image is an image obtained after the to-be-processed image is subjected to deblurring processing; and outputting the processed image. According to the method, the parameter quantity and the calculation quantity of the model can be reduced, and the image deblurring accuracy is guaranteed.

Description

Technical Field

[0001] The present application relates to image processing technology, and in particular to an image processing method, device and readable storage medium. Background Art

[0002] During the process of image capture or transmission, the image becomes blurred due to various factors, such as out-of-focus blur caused by improper focusing during image capture, motion blur caused by the relative movement between the camera and the object being photographed, etc., which causes the image clarity to decrease, seriously affecting the image quality and thus having a negative impact on image analysis, processing and other applications. Therefore, it is necessary to deblur the blurred image to improve its clarity.

[0003] In the prior art, three convolutional neural networks are often connected in series to obtain the deblurred image. However, this method requires the use of three convolutional neural networks in series, and the number of model parameters and the amount of calculation are large. Some studies have proposed using deep separable convolution to replace the convolution method in the convolutional neural network model to reduce the number of model parameters and the amount of calculation. However, while reducing the number of parameters and the amount of calculation, this method will lose some feature information of the image, that is, reduce the accuracy of image deblurring.

[0004] Therefore, a processing method is needed that can both ensure the accuracy of image deblurring and reduce the number of model parameters and the amount of calculation. Summary of the invention

[0005] The present application provides an image processing method, device and readable storage medium, which are used to reduce the number of parameters and calculation amount of a model and ensure the accuracy of image deblurring.

[0006] In a first aspect, the present application provides an image processing method, comprising:

[0007] Get the image to be processed;

[0008] Performing image transformation on the image to be processed to obtain N first images; the size of the first images is the same as that of the image to be processed; and N is an integer greater than or equal to 2;

[0009] Processing the N first images using an autoencoder model to obtain a second image corresponding to each of the first images; the autoencoder model is used to perform deblurring processing on the image;

[0010] Performing an inverse image transformation on each of the second images to obtain a third image corresponding to each of the second images;

[0011] Obtaining a processed image according to each of the third images; the processed image is an image obtained by deblurring the image to be processed;

[0012] The processed image is output.

[0013] Optionally, performing image transformation on the image to be processed to obtain N first images includes:

[0014] According to the type of the image to be processed, obtaining N image transformation modes of the image to be processed;

[0015] According to the N image transformation modes of the image to be processed, the image to be processed is transformed to obtain N first images.

[0016] Optionally, performing an image inverse transformation on each of the second images to obtain a third image corresponding to each of the second images includes:

[0017] Obtaining the order of the N image transformation modes;

[0018] Acquire a sorting order of second images corresponding to the first images, and an order of image inverse transformation methods, wherein the sorting order and the order of the image inverse transformation methods are consistent with an order of N image transformation methods;

[0019] According to the order of the corresponding image inverse transformation methods, the sorted second images are subjected to image inverse transformation to obtain third images corresponding to each of the second images.

[0020] Optionally, the using an autoencoder model to process the N first images to obtain a second image corresponding to each of the first images includes:

[0021] According to the order of the N image transformation modes, the N first images are processed in sequence using a pre-trained autoencoder model to obtain a second image corresponding to each of the first images;

[0022] The obtaining the sorting order of the second images corresponding to the first images includes:

[0023] The order of the second images corresponding to the first images acquired in sequence is used as the sorting order of the second images corresponding to the first images.

[0024] Optionally, the using an autoencoder model to process the N first images to obtain a second image corresponding to each of the first images includes:

[0025] The N first images are processed using N autoencoder models to obtain second images corresponding to each of the first images, and the model structures of the N autoencoder models are the same.

[0026] Optionally, the using the autoencoder model to sequentially process the N first images to obtain a second image corresponding to each of the first images further includes:

[0027] According to the size of the image to be processed, the autoencoder model corresponding to the size of the image to be processed is searched from the model library.

[0028] Optionally, obtaining the image to be processed includes:

[0029] Acquiring the image to be processed;

[0030] Or, receiving the image to be processed sent by an image acquisition device or an electronic device.

[0031] Optionally, the method further includes:

[0032] Perform image recognition on the processed image to obtain the image recognition result.

[0033] In a second aspect, the present application provides an image processing device, comprising:

[0034] A first acquisition module, used for acquiring an image to be processed;

[0035] A transformation module, used for performing image transformation on the image to be processed to obtain N first images; the size of the first images is the same as that of the image to be processed; and N is an integer greater than or equal to 2;

[0036] A second acquisition module, used to process the N first images using an autoencoder model to obtain a second image corresponding to each of the first images; the autoencoder model is used to perform deblurring processing on the image;

[0037] an inverse transformation module, configured to perform an inverse transformation on each of the second images to obtain a third image corresponding to each of the second images;

[0038] A processing module, used for obtaining a processed image according to each of the third images; the processed image is an image obtained by deblurring the image to be processed;

[0039] An output module is used to output the processed image.

[0040] In a third aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0041] The memory stores computer-executable instructions;

[0042] The processor executes the computer-executable instructions stored in the memory to implement the image processing method as described in any one of the first aspects.

[0043] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the image processing method as described in any one of the first aspects.

[0044] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the image processing method as described in any one of the first aspects.

[0045] In a sixth aspect, the present application provides a chip having a computer program stored thereon, wherein when the computer program is executed by the chip, the image processing method as described in any one of the first aspects is implemented.

[0046] The image processing method, device and readable storage medium provided by the present application first perform image transformation on the image to be processed to obtain a first image including multiple angle features, then process the first image using an autoencoder model to obtain a second image, and perform image inverse transformation on the second image to obtain a third image, and finally obtain a processed image based on the third image. The method uses reversible image transformation and an autoencoder model to obtain clear image features of the image to be processed at multiple angles, while not losing image feature information, improving the accuracy of the deblurred image, and only using one autoencoder model for image processing, compared with the series connection of N convolutional neural network models in the prior art, reducing the number of model parameters and the amount of calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0048] Figure 1 A flowchart of an image processing method provided in this application;

[0049] Figure 2 A schematic diagram of the structure of an autoencoder model provided for this application;

[0050] Figure 3 A schematic diagram of another image processing method provided by the present application;

[0051] Figure 4 A flowchart of another image processing method is provided for this application;

[0052] Figure 5 A flowchart of another image processing method provided in this application;

[0053] Figure 6A structural schematic diagram of an image processing method provided in this application;

[0054] Figure 7 A schematic diagram of the structure of an image processing device provided by the present application;

[0055] Figure 8 A schematic diagram of the structure of an electronic device provided in this application.

[0056] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0057] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0058] First, the terms involved in this application are explained:

[0059] Image clarity: The clarity of each detail and its boundary of the image. The blurriness of an image can be used to describe the blurriness of the image, which is inversely proportional to the clarity. Therefore, the clarity of an image can be described by its clarity or blurriness. For example, the blurriness of an image can be obtained through a blurriness evaluation algorithm, and then the clarity of the image can be obtained.

[0060] Blurry image: an image whose clarity is lower than a certain preset threshold. Correspondingly, a clear image refers to an image whose clarity is higher than a certain preset threshold.

[0061] During the process of image capture or transmission, the image becomes blurred due to the influence of various factors, such as defocus blur caused by incorrect focusing when capturing the image, motion blur caused by the relative movement between the camera and the photographed object, etc., resulting in a decrease in image clarity, seriously affecting the quality of the image, and thus having a negative impact on image analysis, processing and other applications.

[0062] In the prior art, a machine learning algorithm is often used to perform image deblurring. The machine learning algorithm is used to learn the features of a blurred image and a clear image pair, and then the machine learning algorithm is used to obtain a clear image of the input blurred image.

[0063] At present, the machine learning algorithm of three convolutional neural networks in series is mostly used to obtain clearer images with higher accuracy. However, this method requires the use of N convolutional neural network models, and the number of model parameters and the amount of calculation are large. Some studies have proposed using deep separable convolution to replace the convolution method in the convolutional neural network model to reduce the number of model parameters and the amount of calculation. However, while reducing the number of parameters and the amount of calculation, this method will lose some feature information of the image, that is, reduce the accuracy of image deblurring.

[0064] In view of this, the present application proposes an image processing method, which uses an autoencoder model to achieve end-to-end feature learning and extraction from blurred images to clear images, and uses image transformation to increase the diversity of image feature information, so that multi-angle image features can be extracted when the autoencoder model is used for deblurring. Without losing image feature information, using only one autoencoder model can reduce the number of model parameters and the amount of calculation.

[0065] The execution subject of this application is an electronic device with processing capabilities, such as a computer, a server, etc.; or an image processing device, which can be implemented by any software and / or hardware, such as a processing chip or a collection of processing chips. It should be noted that the image processing device can be set up separately or integrated in an electronic device.

[0066] The following takes the execution subject as an electronic device as an example, and describes in detail the technical solution of the present application and how the technical solution of the present application solves the above technical problems in combination with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in combination with the accompanying drawings.

[0067] Figure 1 A flowchart of an image processing method provided in this application is shown in FIG. Figure 1 As shown, the method includes:

[0068] S101: Obtain an image to be processed.

[0069] The above-mentioned image to be processed refers to an image whose clarity is lower than a preset threshold, that is, a blurred image. It should be noted that the threshold may be different in image applications in different fields, for example, the thresholds in the field of autonomous driving and the field of medical applications may be different.

[0070] In one possible implementation, the electronic device is provided with an image acquisition device, and the image to be processed may be acquired by the image acquisition device. In another possible implementation, the image to be processed is sent to the electronic device by the image acquisition device, or by other electronic devices. In another possible implementation, the image to be processed may be imported from an external device, such as a USB flash drive. In another possible implementation, the image to be processed may be obtained from the cloud or a database.

[0071] S102: Perform image transformation on the image to be processed to obtain N first images.

[0072] The above-mentioned N is an integer greater than or equal to 2.

[0073] The size of the first image is the same as that of the image to be processed. The first image may include the image to be processed and the image to be processed after undergoing different image transformations, or may include only the image to be processed after undergoing different image transformations.

[0074] The above-mentioned image transformation refers to a two-dimensional linear reversible transformation of the image to be processed. The image transformation may be, for example, translation, rotation, shearing, etc. It should be noted that the image transformation of the image to be processed is reversible without losing the characteristic information of the image to be processed. It should be understood that the above-mentioned image transformation corresponds to respective image transformation functions, or image transformation matrices, and based on the image transformation function or image transformation matrix, the image to be processed is transformed to obtain N first images.

[0075] It should be noted that this application does not limit the type of image transformation and can be set according to actual needs.

[0076] Regarding how to perform image transformation on the image to be processed, one possible implementation method is to process each image to be processed using a universal image transformation method, for example, two image transformation methods of translation and rotation are fixed for the image to be processed.

[0077] In another possible implementation, the electronic device has a preset image transformation library, which includes image transformation methods corresponding to multiple image types. The image transformation method corresponding to the image to be processed can be obtained according to the type of the image to be processed, and the image to be processed can be transformed based on the image transformation method corresponding to the image to be processed.

[0078] S103: Process the N first images using the autoencoder model to obtain a second image corresponding to each first image.

[0079] The above-mentioned autoencoder model is used to perform deblurring processing on an image. The above-mentioned second image refers to an image deblurred by using the autoencoder model.

[0080] In one possible implementation, an autoencoder model is used to process the N first images in a certain order to obtain a second image corresponding to each first image. The order may be, for example, the order of the aforementioned image transformation.

[0081] Another possible implementation method is to use N autoencoder models with the same model structure to process the N first images respectively to obtain a second image corresponding to each first image.

[0082] Figure 2 This is a schematic diagram of the structure of an autoencoder model provided in this application. Figure 2 As shown, the autoencoder model may include an input layer, an encoder, a hidden layer, a decoder, and an output layer.

[0083] The input layer is used to input the image to the encoder. The encoder includes multiple convolution layers (Conv) with a convolution kernel size of 3*3 and a step size of 2 (s=2), which are used to extract image feature information of the input image. The hidden layer includes a flattening layer, a reorganization layer and at least one fully connected layer, which is used to perform feature fusion on the extracted image feature information.

[0084] The decoder includes multiple layers of deconvolution layers with a convolution kernel size of 3*3 and a step size of 2 and a deconvolution layer (Deconv) with a convolution kernel size of 3*3 and a step size of 1, which are used to restore the image feature information after feature fusion to the size of the input image. The output layer is used to output the image obtained by the decoder.

[0085] It should be noted that the structure of the above autoencoder model is related to the image size of the input layer of the model.

[0086] S104: Perform an inverse image transformation on each second image to obtain a third image corresponding to each second image.

[0087] The third image is the second image after the inverse transformation. As mentioned above, the image transformation is a reversible transformation. To ensure the accuracy of the deblurred image, the second image processed by the autoencoder model needs to be inversely transformed.

[0088] It should be understood that the above-mentioned inverse image transformations correspond to respective inverse image transformation functions, or inverse image transformation matrices. Based on the inverse image transformation functions, or inverse image transformation matrices, the above-mentioned second images are inversely transformed to obtain third images corresponding to each second image.

[0089] S105: Obtain a processed image according to each third image.

[0090] The processed image is an image obtained by deblurring the image to be processed.

[0091] One possible implementation method is to average the data of the same pixel point in the above-mentioned third images to obtain the average value of the pixel point, and use the average value of each pixel point as the data of the corresponding pixel point of the processed image, or multiply the average value of each pixel point by a coefficient as the data of the corresponding pixel point of the processed image.

[0092] Another possible implementation method is to preset weight values ​​of second images of different scales, and the third images corresponding to the second images have the same weight value. Based on the different weight values, the data of the same pixel point of each third image is weighted averaged to obtain the average value of the pixel point, and the average value of the pixel point is used as the data of the pixel point of the processed image, or the value obtained by multiplying the average value of each pixel point by a coefficient is used as the data of the corresponding pixel point of the processed image.

[0093] S106: Output the processed image.

[0094] In a possible implementation, the electronic device includes an image recognition device, and the processed image is output to the image recognition device of the electronic device for image recognition to obtain an image recognition result for subsequent image application. For example, autonomous driving is performed based on the image recognition result.

[0095] In another possible implementation, the electronic device outputs the deblurred image to an image recognition device for image recognition to obtain image recognition results for subsequent image applications. The image recognition device is independent of the electronic device and can be implemented by any software and / or hardware, such as a processing chip or a collection of processing chips.

[0096] The image processing method provided by the present application first performs an image transformation on the image to be processed to obtain a first image including features of multiple angles, then processes the first image using an autoencoder model to obtain a second image, and performs an image inverse transformation on the second image to obtain a third image, and finally obtains a processed image based on the third image. The method uses reversible image transformation and an autoencoder model to obtain clear image features of the image to be processed at multiple angles, while not losing image feature information, improving the accuracy of the deblurred image, and only using one autoencoder model for image processing, compared with the stacking of N convolutional neural network models in the prior art, reducing the number of model parameters and the amount of calculation.

[0097] The following takes the above electronic device including the image transformation library and the model library as an example to explain how to perform image transformation on the image to be processed to obtain N first images.

[0098] Figure 3 A schematic diagram of another image processing method provided in this application. Figure 3 As shown, the method includes:

[0099] S201. Obtain N image transformation modes of the image to be processed according to the type of the image to be processed.

[0100] Exemplarily, the image transformation library included in the electronic device includes image transformation methods corresponding to multiple image types. The image type can refer to the format of the picture, such as bmp, jpg, tif, etc.; it can also be the type of image, such as grayscale image, true color RGB image, indexed image, etc., which can be set according to actual needs. Table 1 is a schematic diagram of an image transformation library.

[0101] Table 1 Image transformation library

[0102] Image Type Image transformation method tif Translation, rotation, shearing jpg Translation, rotation

[0103] S202 . Perform image transformation on the image to be processed according to N image transformation modes of the image to be processed to obtain N first images.

[0104] It should be understood that the image data is essentially a pixel data matrix. Taking the image transformation method as translation as an example, the pixel data matrix of the image to be processed is multiplied by the translation matrix to obtain the first image. Other image transformation methods are similar to this principle and will not be described in detail here.

[0105] S203. According to the size of the image to be processed, search the model library for an autoencoder model corresponding to the size of the image to be processed.

[0106] The above model library includes pre-trained autoencoder models of various scales. As mentioned above, the structure of the above autoencoder model is related to the image size of the input layer of the model, so the autoencoder model corresponding to the size of the image to be processed can be found from the model library according to the size of the image to be processed.

[0107] S204: Process the N first images using the autoencoder model to obtain a second image corresponding to each first image.

[0108] The image processing method provided in the present application can determine the image transformation method and autoencoder model of the image to be processed based on the image to be processed, and can determine how to process the image to be processed based on the properties of the image itself, so that the acquired first image and second image are more accurate.

[0109] The following describes how to perform image inverse transformation on each second image to obtain a third image corresponding to each second image.

[0110] Figure 4 A schematic diagram of another image processing method provided in this application. Figure 4 As shown, the method includes:

[0111] S301: Obtain the order of N image transformation modes.

[0112] Regarding the image transformation method, a possible implementation method is that the above-mentioned image transformation library can also include the order of the image transformation methods. For example, refer to Table 1. The position of the name of the image transformation in Table 1 is the order of the image transformation method.

[0113] In another possible implementation, the order of all image transformation methods is preset. For example, if the image transformation methods include translation, rotation, and shearing, the translation is preset first, then the rotation, and finally the shearing. Regardless of the type of image transformation method corresponding to the image to be processed, the image transformation is performed in the preset order, and the method that does not exist in the image transformation method corresponding to the image to be processed is not processed.

[0114] In another possible implementation, the above-mentioned image transformation method of the image to be processed is universal, and the order of the image transformation method is also fixed.

[0115] S302. According to the order of the N image transformation methods, the N first images are processed in sequence using a pre-trained autoencoder model to obtain a second image corresponding to each first image.

[0116] Due to the order of the image transformation method, N first images will be obtained in sequence. Therefore, the first images can be processed in sequence using the autoencoder model according to the order of the image transformation method to obtain second images corresponding to each first image.

[0117] S303: Obtain the sorting order of the second images corresponding to the first images, and the order of the image inverse transformation methods.

[0118] The above sorting order and the order of the image inverse transformation method are consistent with the order of the N image transformation methods.

[0119] Exemplarily, the order of the second images corresponding to the first images acquired sequentially is used as the sorting order of the second images corresponding to the first images.

[0120] S304 . Perform image inverse transformation on the sorted second images in the order of the corresponding image inverse transformation methods to obtain third images corresponding to the second images.

[0121] This step can ensure the accuracy of the correspondence between the image transformation and the image inverse transformation. If the correspondence between the image inverse transformation and the image transformation is disordered, it will lead to errors in image features, and thus cause the final deblurred image to be inaccurate.

[0122] The image processing method provided by the present application performs image transformation processing, deblurring processing, and image inverse transformation processing on the image to be processed in sequence based on the order of the image transformation method, and finally obtains a third image. This method can ensure the corresponding relationship between image transformation and image inverse transformation and improve the accuracy of image processing.

[0123] The image processing method provided in the present application is described below by taking the use of N autoencoder models to process N first images to obtain a second image corresponding to each first image as an example.

[0124] Figure 5 A flowchart of another image processing method provided in this application. Figure 6 This is a schematic diagram of the structure of an image processing method provided in this application. It should be noted that: Figure 6 The example is just to illustrate the correspondence between the to-be-processed images and the two first images. Figure 5 and Figure 6 As shown, the method includes:

[0125] S401: Obtain an image to be processed.

[0126] S402 , performing image transformation on the image to be processed to obtain N first images.

[0127] This step can refer to the aforementioned steps S201-202, which will not be described in detail here. Figure 6 After image transformation, the first image-1 and the first image-2 can be obtained.

[0128] S403: Use N autoencoder models to process the N first images to obtain a second image corresponding to each first image.

[0129] Exemplarily, this step can use N autoencoder models to process N first images in parallel to obtain the second image corresponding to each first image. Figure 6 , for the first image-1 and the first image-2, two autoencoder models with the same model structure are used in parallel to obtain the second image-1 and the second image-2 respectively.

[0130] S404: Perform an inverse image transformation on each second image to obtain a third image corresponding to each second image.

[0131] This step may refer to the aforementioned steps S303 - 304 , which will not be described in detail here.

[0132] S405: Obtain a processed image according to each third image.

[0133] S406: Output the processed image.

[0134] The image processing method provided in the present application can utilize N autoencoder models to process N first images in parallel, and obtain N second images corresponding to the N first images, thereby improving the efficiency of image processing.

[0135] Figure 7 This is a schematic diagram of the structure of an image processing device provided by this application. Figure 7 As shown, the device comprises:

[0136] A first acquisition module 11, used to acquire an image to be processed;

[0137] A transformation module 12, configured to perform image transformation on the image to be processed to obtain N first images; the size of the first images is the same as that of the image to be processed; and N is an integer greater than or equal to 2;

[0138] A second acquisition module 13 is used to process the N first images using an autoencoder model to obtain a second image corresponding to each of the first images; the autoencoder model is used to perform deblurring processing on the image;

[0139] An inverse transformation module 14, configured to perform an image inverse transformation on each of the second images to obtain a third image corresponding to each of the second images;

[0140] A processing module 15 is used to obtain a processed image according to each of the third images; the processed image is an image obtained by deblurring the image to be processed;

[0141] The output module 16 is used to output the processed image.

[0142] In a possible implementation, the transformation module 12 is specifically used to obtain N image transformation modes of the image to be processed according to the type of the image to be processed; and perform image transformation on the image to be processed according to the N image transformation modes of the image to be processed to obtain N first images.

[0143] In a possible implementation, the inverse transformation module 14 is specifically used to obtain the order of the N image transformation methods; obtain the sorting order of the second images corresponding to the first image, and the order of the image inverse transformation methods, and the sorting order and the order of the image inverse transformation methods are consistent with the order of the N image transformation methods; perform image inverse transformation on the sorted second images according to the order of the corresponding image inverse transformation methods to obtain third images corresponding to each of the second images.

[0144] In a possible implementation method, the second acquisition module 13 is specifically used to process the N first images in sequence according to the order of the N image transformation methods using a pre-trained autoencoder model to obtain a second image corresponding to each of the first images; and the inverse transformation module 14 is specifically used to use the order of the second images corresponding to the first images obtained in sequence as the sorting order of the second images corresponding to the first images.

[0145] In a possible implementation, the second acquisition module 13 is specifically used to process the N first images using N autoencoder models to obtain a second image corresponding to each of the first images, and the model structures of the N autoencoder models are the same.

[0146] In a possible implementation, the second acquisition module 13 is specifically used to search the autoencoder model corresponding to the size of the image to be processed from the model library according to the size of the image to be processed.

[0147] In a possible implementation, the first acquisition module 11 is specifically configured to acquire the image to be processed; or receive the image to be processed sent by an image acquisition device or an electronic device.

[0148] In a possible implementation, the recognition module 17 is specifically configured to perform image recognition on the processed image to obtain the image recognition result.

[0149] The functional modules of the above-mentioned image processing device can cooperate to execute the image processing method described in the above-mentioned embodiment. The implementation principle and technical effect are similar and will not be described in detail.

[0150] Figure 8 This is a schematic diagram of the structure of an electronic device provided in this application. Figure 8 As shown, the electronic device 500 may include: at least one processor 501 and a memory 502. The electronic device may be a terminal, a server, a computer device, or other device with processing capabilities.

[0151] The memory 502 is used to store programs. Specifically, the programs may include program codes, and the program codes include computer operation instructions.

[0152] The memory 502 may include a high-speed RAM memory, and may also include a non-volatile memory.

[0153] The processor 501 is used to execute the computer-executable instructions stored in the memory 502 to implement the image processing method described in the above method embodiment. The processor 501 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0154] The electronic device 500 may further include a communication interface 503, so as to communicate and interact with an external device through the communication interface 503. The external device may be, for example, a computer, a tablet, and the like.

[0155] In specific implementation, if the communication interface 503, the memory 502 and the processor 501 are implemented independently, the communication interface 503, the memory 502 and the processor 501 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc., but it does not mean that there is only one bus or one type of bus.

[0156] Optionally, in a specific implementation, if the communication interface 503, the memory 502 and the processor 501 are integrated on a chip, the communication interface 503, the memory 502 and the processor 501 can communicate through an internal interface.

[0157] The present application also provides a computer-readable storage medium, which may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, and other media that can store program codes. Specifically, the computer-readable storage medium stores computer execution instructions, and the computer execution instructions are used for the image processing method in the above-mentioned embodiment.

[0158] The present application also provides a computer program product, which includes an execution instruction, which is stored in a readable storage medium. At least one processor of the electronic device 500 can read the execution instruction from the readable storage medium, and at least one processor executes the execution instruction so that the electronic device 500 implements the methods provided in the various embodiments described above.

[0159] The present application also provides a chip having a computer program stored thereon, and when the computer program is executed by the chip, the methods provided in various implementation modes are implemented.

[0160] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0161] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. An image processing method, characterized in that: The method comprises: Get the image to be processed; Performing image transformation on the image to be processed to obtain N first images; the size of the first images is the same as that of the image to be processed; and N is an integer greater than or equal to 2; Processing the N first images using an autoencoder model to obtain a second image corresponding to each of the first images; the autoencoder model is used to perform deblurring processing on the image; Performing an inverse image transformation on each of the second images to obtain a third image corresponding to each of the second images; Obtaining a processed image according to each of the third images; the processed image is an image obtained by deblurring the image to be processed; The processed image is output.

2. The method according to claim 1, characterized in that The performing image transformation on the image to be processed to obtain N first images includes: According to the type of the image to be processed, obtaining N image transformation modes of the image to be processed; According to the N image transformation modes of the image to be processed, the image to be processed is transformed to obtain N first images.

3. The method according to claim 2, characterized in that The performing an image inverse transformation on each of the second images to obtain a third image corresponding to each of the second images includes: Obtaining the order of the N image transformation modes; Acquire a sorting order of second images corresponding to the first images, and an order of image inverse transformation methods, wherein the sorting order and the order of the image inverse transformation methods are consistent with an order of N image transformation methods; According to the order of the corresponding image inverse transformation methods, the sorted second images are subjected to image inverse transformation to obtain third images corresponding to each of the second images.

4. The method according to claim 3, characterized in that The using the autoencoder model to process the N first images to obtain a second image corresponding to each of the first images includes: According to the order of the N image transformation modes, the N first images are processed in sequence using a pre-trained autoencoder model to obtain a second image corresponding to each of the first images; The obtaining the sorting order of the second images corresponding to the first images includes: The order of the second images corresponding to the first images acquired in sequence is used as the sorting order of the second images corresponding to the first images.

5. The method according to any one of claims 1 to 3, characterized in that: The using the autoencoder model to process the N first images to obtain a second image corresponding to each of the first images includes: The N first images are processed using N autoencoder models to obtain second images corresponding to each of the first images, and the model structures of the N autoencoder models are the same.

6. The method according to any one of claims 1 to 3, characterized in that: Before sequentially processing the N first images using the autoencoder model to obtain the second image corresponding to each of the first images, the method further includes: According to the size of the image to be processed, the autoencoder model corresponding to the size of the image to be processed is searched from the model library.

7. The method according to any one of claims 1 to 3, characterized in that: The step of obtaining the image to be processed comprises: Acquiring the image to be processed; Or, receiving the image to be processed sent by an image acquisition device or an electronic device.

8. The method according to any one of claims 1 to 3, characterized in that: The method further comprises: Perform image recognition on the processed image to obtain an image recognition result.

9. An electronic device, characterized in that: The electronic device comprises: a processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the image processing method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the image processing method according to any one of claims 1 to 8 when executed by a processor.

Images