CN112053285A

CN112053285A - Image processing method, image processing device, computer equipment and storage medium

Info

Publication number: CN112053285A
Application number: CN202010905635.1A
Authority: CN
Inventors: 陈嘉莉; 周超勇; 刘玉宇
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2020-09-01
Filing date: 2020-09-01
Publication date: 2020-12-08
Anticipated expiration: 2040-09-01
Also published as: CN112053285B

Abstract

The invention relates to the field of artificial intelligence, and provides an image processing method, an image processing device, computer equipment and a storage medium. According to the image processing method, the image processing device, the computer equipment and the storage medium, after the image to be processed is obtained, the image to be processed is processed by the residual error encoder and the image encoder respectively to generate residual error image information and coded image information, more texture features of the image to be processed can be reserved through the residual error encoder, so that the features of the image are enriched, a target image with higher resolution is generated after an image decoder decodes the target image, and the quality of the target image is improved.

Description

Image processing method, image processing device, computer equipment and storage medium

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to an image processing method and apparatus, a computer device, and a storage medium.

Background

With the development of digitization, the requirements for image Resolution are also increased, for example, the existing Super-Resolution technology (SR) can reconstruct a corresponding high-Resolution image from an observed low-Resolution image, and has important application values in the fields of monitoring equipment, satellite images, medical images and the like. The Super-Resolution technology based on deep learning is mainly a reconstruction method based on a Single low-Resolution Image, namely (Single Image Super-Resolution, SISR), and the existing Super-Resolution technology often has errors and the like in the Image processing process, so that the problem of low quality is easily caused when the low-Resolution Image is converted into the high-Resolution Image. Therefore, how to obtain high-resolution images with higher quality becomes a problem to be solved by those skilled in the art.

Disclosure of Invention

The present invention is directed to an image processing method, an image processing apparatus, a computer device, and a storage medium, so as to solve the problem of poor quality when converting a low-resolution image into a high-resolution image in the prior art.

In order to achieve the above object, the present invention provides an image processing method comprising:

acquiring an image to be processed;

processing the image to be processed by adopting a residual encoder to acquire residual image information;

converting the image to be processed into coded image information by adopting an image coder;

and decoding by adopting an image decoder according to the residual image information and the coded image information to obtain a target image.

Further, the processing the image to be processed by using the residual encoder to obtain residual image information includes:

obtaining a pixel value of the image to be processed, and subtracting the pixel value of a preset image from the pixel value of the image to be processed to obtain residual image data;

and encoding the residual image data by adopting the residual encoder to obtain the residual image information.

Further, the image to be processed is converted into encoded image information by using an image encoder, including;

and the image encoder performs feature extraction on the image to be processed to obtain image features, and converts the image features into encoded image information.

Further, before the decoding is performed by the image decoder according to the residual image information and the encoded image information, and the target image is obtained, the method may further include:

and performing noise reduction processing on the coded image information by adopting a bilateral filtering algorithm.

Further, the decoding by using an image decoder according to the residual image information and the encoded image information to obtain a target image includes:

and the image decoder combines the residual image information and corresponding information in the coded image information, and sequentially performs entropy decoding and inverse quantization processing to acquire the target image.

Further, after acquiring the image to be processed, the method further includes:

and cutting the image to be processed, and performing noise reduction on the cut image to be processed by adopting an Adam algorithm to obtain the image to be processed after noise reduction.

In order to achieve the above object, the present invention provides an image processing apparatus, which includes an obtaining module, a residual coding module, an image coding module and a decoding module, wherein the obtaining module is configured to obtain an image to be processed, the residual coding module is configured to process the image to be processed by using a residual coder to obtain residual image information, the image coding module is configured to convert the image to be processed into coded image information by using an image coder, and the decoding module is configured to decode the coded image information by using an image decoder according to the residual image information and the coded image information to obtain a target image.

Further, the residual coding module obtains a pixel value of the image to be processed, subtracts a pixel value of a preset image from the pixel value of the image to be processed to obtain residual image data, and codes the residual image data by using the residual coder to obtain the residual image information.

In order to achieve the above object, the present invention also provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the aforementioned method when executing the computer program.

In order to achieve the above object, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which when executed by a processor implements the steps of the aforementioned method.

The image processing method, the image processing device, the computer equipment and the storage medium respectively process the image to be processed by the residual error encoder and the image encoder after the image to be processed is obtained, so that residual error image information and coded image information are generated, more texture characteristics of the image to be processed can be reserved through the residual error encoder, the characteristics of the image are enriched, a target image with higher resolution is generated after an image decoder decodes the target image, and the quality of the output target image is improved because the image characteristics are reserved through the residual error encoder.

Drawings

FIG. 1 is a block flow diagram of one embodiment of an image processing method of the present invention;

FIG. 2 is a diagram illustrating an embodiment of an image processing method according to the present invention;

FIG. 3 is a block diagram of an embodiment of an image processing apparatus according to the present invention;

fig. 4 is a hardware architecture diagram of one embodiment of the computer apparatus of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The image processing method, the image processing device, the computer equipment and the storage medium are suitable for the field of intelligent medical treatment. The invention acquires the image to be processed, processes the image to be processed by the residual encoder to acquire the residual image information, can better reserve the texture characteristic after the image processing through the residual image information, converts the image to be processed into the encoded image data by the image encoder, and respectively processes the image to be processed through two different codes, thereby enriching the characteristic of the image information.

Example one

Referring to fig. 1 and 2, an image processing method of the present embodiment is shown, which includes the following steps:

s1, acquiring an image to be processed;

the image to be processed may be an image with a lower resolution, the image to be processed may be a three-channel RGB image, the three-channel RGB may be used to extract image data of the image to be processed, for example, an image with a resolution of 640 × 480 is used as the image to be processed, an existing arbitrary image may be selected as the image to be processed, each pixel point in the image to be processed may be represented by three values, so the image to be processed may also be referred to as a three-channel image, an RGB color mode is a color standard in the industry, and various colors are obtained by changing three color channels of red (R), green (G), and blue (B) and superimposing the three color channels with each other, and RGB represents colors of the three channels of red, green, and blue.

S2, processing the image to be processed by adopting a residual encoder to obtain residual image information;

adding residual details in the calculation by a residual encoder, specifically, the method includes: the method comprises the steps of obtaining pixel values of an image to be processed, wherein each pixel value is a data parameter representing a corresponding pixel point, one image can be composed of all the pixel values of the image, subtracting the pixel value of a preset image from the pixel value of the image to be processed, the preset image generally refers to expected or ideal image data so as to obtain residual image data, the residual image data contains residual details, and encoding the residual image data by using a residual encoder so as to obtain residual image information. More texture features of the image to be processed can be reserved through the residual encoder processing, so that the features of the image are enriched, and a target image with higher resolution is generated after the image decoder decodes the target image.

In this embodiment, the residual encoding by the residual encoder may include: carrying out convolution processing on the image to be processed for multiple times, and adopting a 3 multiplied by 3 convolution kernel; at each convolution processing, the residual values of the image to be processed are added. When the image features of the image to be processed are extracted through multiple convolutions, when the residual encoder performs residual encoding, a shortcut is added between data of every two layers (data processing), each layer represents a plurality of data processing processes on a certain level or gradient to form a residual block, the residual encoding is easier to be optimized by adding shortcuts connections, for a network (data processing) of several layers containing one shortcut connection, the network (data processing) is called a residual block (residual block), for example, x represents input, f (x) represents output of the residual block before an activation function of a second layer, and f (x) represents W2 σ (W1x), wherein W1 represents weight of the first layer, W2 represents weight of the second layer (i.e., before and after each convolution processing), and σ represents a ReLU activation function. The output of the last residual block is σ (F (x) + x). When there is no shortcut connection, the residual block is a normal two-layer network. The network in the residual block may be a fully connected layer or a convolutional layer.

Assume the output h (x) of the layer two network before the activation function. If the optimal output is input x in the two-layer network, then for a network without shortcut connection, it needs to be optimized to h (x) ═ x; for a network with short connection, i.e., a residual block structure, if the optimal output is x, f (x) ═ h (x) — x only needs to be optimized to 0. And a plurality of residual blocks are connected together to form a residual network, and the output characteristic diagram before downsampling of each convolution processing is spliced to the input characteristic diagram of the decoder of the corresponding channel, so that information transmission of a plurality of levels is realized.

The residual block can be optimized in a calculation mode, a high-dimensional convolutional layer structure is replaced by a low-dimensional convolutional layer structure, two 3 x 3 convolutional layers can be replaced by (1 x 1, 3 x 3, 1x 1) convolutional layers, then reduction is carried out under the other 1x 1 convolutional layer, and the method can be applied to dimension reduction processing of other dimensions in the same way, so that the precision is maintained, the calculation amount is reduced, and the calculation cost is reduced. The residual error network can be trained in a string, and divided into individual residual error blocks for training, so that the error of each residual error block is minimum, and finally the purpose of minimum overall error is achieved, and the gradient dispersion phenomenon is avoided.

In order to compress the features and simplify the complexity of calculation, the image to be processed is subjected to pooling after convolution processing, and the pooling includes average pooling and maximum pooling. Pooling can be used for extracting more expensive features in one step, and some detail information is omitted through pooling processing to achieve down-sampling, so that the calculated amount of image processing is reduced, the outstanding background features can be better reserved through average pooling, and the outstanding texture features can be better reserved through maximum pooling. Maximum pooling extracts a feature in one quadrant, which is kept in the maximized pooled output, and keeps its maximum; the average pooling is to average the extracted values in one quadrant, thus ensuring that the sum of the gradients before and after pooling remains unchanged.

S3, converting the image to be processed into coded image information by adopting an image coder;

in the present embodiment, converting the image to be processed into the encoded image information includes: the image to be processed is subjected to feature extraction, a convolution processing mode can be adopted to obtain image features, the feature extraction is realized through convolution, and then the image features are converted into coded image information.

In a specific implementation process, the convolution processing includes a plurality of convolution layers, that is, a plurality of convolution calculations are performed, and features of the image to be processed are gradually extracted through convolution by using a 3 × 3 convolution kernel. If the input image is a 572 × 572 × 1 image, the input image is convolved by a (3 × 3 convolution kernel) and converted to 570 × 570 × 64. Each convolution process generates many small patches of feature maps or features that preserve the relationship between pixels in the input image.

And S4, decoding by adopting an image decoder according to the residual image information and the coded image information to obtain a target image.

In order to improve the image quality, the method further comprises the following steps before the target image is acquired: the coded image information is subjected to noise reduction by adopting a Bilateral filtering algorithm, a low-quality full-resolution image is subjected to enhancement processing by adopting a Bilateral filtering (Bilateral filter) algorithm so as to achieve the purposes of removing image particle noise and keeping image edges and detail textures, a human visual attention mechanism is sensitive to the information such as the edge textures, and the subjective visual quality of the image can be improved after the Bilateral filtering algorithm is adopted for enhancement. The bilateral filtering algorithm is a classical filtering enhancement algorithm, is a nonlinear filtering method, is a compromise treatment combining the spatial proximity and the pixel value similarity of an image, simultaneously considers the spatial domain information and the gray level similarity, achieves the purpose of edge-preserving and denoising, and has the characteristics of simplicity, non-iteration and locality.

Specifically, when the target image is obtained, the image decoder merges residual image information and corresponding information in the encoded image information, where the corresponding information includes corresponding image features, and the image features may include color features, texture features, shape features, spatial relationship features, and the like, and the merging includes integrating (fusing) image data on the image features to generate merged information, and then decoding image data of the same feature on a corresponding channel by transpose convolution (deconvolution). And after the information data is combined, entropy decoding and inverse quantization are sequentially carried out on the information data to obtain the target image, so that the information data is further optimized, and the quality of information processing is improved. Entropy coding is to use codes with different lengths to represent different characters according to the occurrence probability of different image data in image information, to use shorter codes to represent image data with higher occurrence probability, and to use longer codes to represent image data with low occurrence probability, thereby improving coding efficiency and reducing storage requirements, and specifically can be implemented by using huffman coding. The inverse quantization is based on the temporary result after Discrete Cosine Transform (DCT) transformation, the temporary result is divided by the respective quantization step length and rounded up to obtain the quantization coefficient, the integer Discrete Cosine Transform is performed through the formed quantization matrix, so that the data information processing is efficient and accurate, and the corresponding matrix operation is reduced.

Generating a target image by transpose convolution decoding, extracting feature information in the image by using continuous convolution pooling layers, gradually mapping the image feature information to a high dimension, the highest dimension of the whole network is rich characteristic information in the whole image, in order to enhance the segmentation precision in the mapping process, the images with the same dimension in the contraction network under the same dimension are merged, because the dimension can become a multiple of the original dimension in the merging process, the dimension needs to be reduced by convolution again at the moment, the dimension after processing is ensured to be the same as the dimension before merging operation, so that the image can be merged with the image under the same dimensionality for the second time after the transposition convolution is carried out for the second time, the features of different layers are grabbed and integrated in a feature merging mode until the image can be output finally when the dimensionality of the original image is the same, for example, the low-resolution image to be processed of 3-channel RGB is output as a high-resolution target image of 3-channel RGB.

In the embodiment, a contraction path is adopted during encoding, and a convolution network structure can be adopted, wherein the contraction path comprises the use of repeated 3 × 3 convolution, the number of characteristic channels can be doubled by sampling each time, and the two modes of a residual error encoder and an image encoder are respectively adopted for encoding; and finally, mapping the component feature vectors corresponding to each pixel to a class with an expected number to generate a target image.

The number of channels of the residual encoder and the image encoder during encoding can be 32, 64, 128, 256 and 512 respectively, the number of channels of the image decoder during decoding can be 256, 128, 64, 32 and 12 respectively, 3 x 3 convolution can be adopted for encoding and decoding, 2 x 2 maximum pooling is used for encoding downsampling, and 2 x 2 transposition convolution is used for decoding upsampling. The convolutions of the residual encoder and the image encoder are combined together according to the corresponding image characteristics and input to the decoder. And the output characteristic information of the residual encoder before each downsampling is merged on the input characteristic information of the image decoder of the corresponding channel to realize information transmission of multiple levels, the decoding process is merged with the corresponding image characteristic extraction part, and the output characteristic information can be cut as required before merging.

For further optimizing the image, after performing step S1, the method may further include: and cutting the image to be processed, and performing noise reduction processing on the cut image to be processed by adopting an Adam algorithm to obtain the image to be processed after noise reduction.

An Adam algorithm is used to perform optimized noise reduction on an image to be processed, and by way of example and not limitation, the parameters are configured as follows: β 1 is 0.9, β 2 is 0.999, E is 1E-8, the learning rate is 1E-4, the learning rate decays after 200epoch, the decay coefficient is 0.5, wherein β 1 is an exponential decay rate, which is called a first moment, β 2 is used to calculate an exponentially weighted average of a square number, which is called a second moment, and E is a natural constant. Independent adaptive learning rates are designed for different parameters by calculating first and second moment estimates of the gradient, and training may be stopped at the 400 th epoch in a particular scenario. The training image may be trained using a high resolution-low resolution image pair, such as now an image with a resolution of 1024 × 2048, cropped to a 256 × 256 patch, to enhance the original image. The Adam algorithm is an algorithm for executing first-order gradient optimization on a random objective function, is based on adaptive low-order moment estimation, has high calculation efficiency and low memory requirement, and is suitable for solving the unsteady-state (non-stationary) problems of large noise and sparse gradient.

The image processing can reduce the video memory and the calculation amount, so that the occupied memory is reduced, the calculation amount is reduced, the same convolution of 3 x 3 can extract the features in a larger image range, the image segmentation and splicing processing is facilitated, and the fusion of multi-scale features can be conveniently carried out through different coding modes. The embodiment can increase feature extraction and realize multi-scale feature fusion, and ensures that all scale information is fully transformed.

The method can divide the coding process into two parts on the basis of the UNet architecture, residual details to be added are calculated through a residual coder, the image content is coded through an image coder, the original image is enhanced through residual coding, therefore, the low-resolution image to be processed can be generated into a high-resolution target image, the information transmission of multiple levels is realized through the splitting coding process, fewer parameter settings can be realized at the same level, and the capacity and the efficiency of image processing are improved. The embodiment increases the size of the parameter by dividing the original encoder into two sub-encoders, so the performance is also improved, the size and the inference time of the incremental parameter are far less than twice of the width of each layer of the UNet architecture, but the performance is improved more, and the generation of high-quality images is realized.

Example two

Referring to fig. 3, an image processing apparatus 10 of the present invention is shown, including an obtaining module 11, a residual coding module 12, an image coding module 13, and a decoding module 14, where the obtaining module 11 is configured to obtain an image to be processed, the residual coding module 12 is configured to process the image to be processed by using a residual encoder to obtain residual image information, the image coding module 13 is configured to convert the image to be processed into coded image information by using an image encoder, and the decoding module 14 is configured to decode the coded image information by using an image decoder according to the residual image information and the coded image information to obtain a target image.

In this embodiment, the obtaining module 11 obtains an image to be processed after the image is acquired, the image to be processed includes image data information of the image to be processed, the residual encoder performs residual encoding in the residual encoding module 12, the residual encoder adds residual details to obtain residual image information, the image encoding module 13 converts the image to be processed into encoded image information, and the decoding module 14 decodes the image to be processed according to the residual image information and the encoded image information by using an image decoder to generate a target image.

Adding residual details in the calculation by the residual encoder in the residual encoding module 12 specifically includes: the method comprises the steps of obtaining pixel values of an image to be processed, wherein each pixel value is a data parameter representing a corresponding pixel point, one image can be composed of all the pixel values of the image, subtracting the pixel value of a preset image from the pixel value of the image to be processed, the preset image generally refers to expected or ideal image data so as to obtain residual image data, the residual image data contains residual details, and encoding the residual image data by using a residual encoder so as to obtain residual image information.

When encoding, the image to be processed may be subjected to convolution processing for multiple times, a 3 × 3 convolution kernel is adopted, image features may be extracted through the convolution processing, the encoding module 12 includes a residual encoder for performing residual encoding, and residual details are added during residual encoding, and the residual details are added through a residual function f (x) ═ h (x) — x, so that texture features after image processing can be better retained. Assuming that the input of certain image data is x, the desired output is h (x), that is, h (x) is a desired mapping, in residual coding, the input x is directly passed to the output as an initial result by means of short connections, and the output result is h (x) ═ f (x) + x, and when f (x) ═ 0, then h (x) ═ x is an identity mapping. Residual coding is the difference between target values h (x) and x, i.e. so-called residual f (x) ═ h (x) — x, and is to approach the residual result to 0 and ensure that the accuracy does not decrease.

The decoding module 14 combines the residual image information with corresponding information in the encoded image information, for example, according to corresponding image features, the image features may include color features, texture features, shape features, spatial relationship features, and the like, and the combination refers to integration (fusion) of image data on the image features to generate combined information, and then the combined information is transposed convolution (deconvolution) to obtain a target image, and the image data of the same feature is decoded on a corresponding channel, and finally restored to the target image.

The decoding of the decoding module 14 generates a target image through transposed convolution, the feature information in the image can be extracted by using a continuous convolution pooling layer, and the image feature information is mapped to a high dimension step by step, the highest dimension of the whole network is rich feature information in the whole image, in order to enhance the segmentation precision in the mapping process, the images with the same dimension in the contraction network under the same dimension can be merged, because the dimension can become a multiple of the original dimension in the merging process, at this time, the dimension needs to be convolved again for dimension reduction, the processed dimension is ensured to be the same as the dimension before merging operation, so that the images with the same dimension can be merged for the second time after the next transposed convolution, the features with different layers are captured, and the features are integrated in a feature merging manner until the images with the original image can be finally output, for example, the image to be processed with low resolution of 3 channels is output as the target image with high resolution of 3 channels RGB Like this.

In order to further optimize an image, the present embodiment may further include a noise reduction module, where the noise reduction module cuts the image to be processed, and performs noise reduction on the cut image to be processed by using an Adam algorithm to obtain the image to be processed after noise reduction. Adopting Adam algorithm to carry out optimized noise reduction on the image to be processed, wherein the parameters are configured as follows: β 1-0.9, β 2-0.999, E-1E-8, learning rate 1E-4, attenuation after 200epoch, and attenuation coefficient 0.5. The scheme can be applied to scenes of intelligent medical treatment, images shot by medical equipment can be processed, and the scheme can also be applied to scenes of monitoring equipment, satellite images, medical images and the like to solve the problem that high-resolution images cannot be obtained due to various limiting factors so as to obtain clearer images and promote the construction of intelligent cities.

EXAMPLE III

The embodiment further provides a computer device 20, components of the image processing apparatus 10 of the second embodiment may be disposed in the computer device 20, and the computer device 20 may be a smart phone, a tablet computer, a notebook computer, a desktop computer, a rack server, a blade server, a tower server, or a rack server (including an independent server or a server cluster formed by multiple servers) for executing programs, and the like. The computer device 20 of the present embodiment includes at least, but is not limited to: a memory 21, a processor 22 and a computer program stored on the memory 21 and executable on the processor 22, which are communicatively connected to each other via a system bus, as shown in fig. 4. It is noted that fig. 4 only shows the computer device 20 with components 21-22, but it is to be understood that not all shown components are required to be implemented, and that more or fewer components may be implemented instead.

In the present embodiment, the memory 21 (i.e., a readable storage medium) includes a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, and the like. In some embodiments, the storage 21 may be an internal storage unit of the computer device 20, such as a hard disk or a memory of the computer device 20. In other embodiments, the memory 21 may also be an external storage device of the computer device 20, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), or the like, provided on the computer device 20. Of course, the memory 21 may also include both internal and external storage devices of the computer device 20. In this embodiment, the memory 21 is generally used for storing an operating system and various application codes installed in the computer device. Further, the memory 21 may also be used to temporarily store various types of data that have been output or are to be output.

Processor 22 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data Processing chip in some embodiments. The processor 22 is typically used to control the overall operation of the computer device 20. In this embodiment, the processor 22 is configured to execute the program code stored in the memory 21 or process data. The image processing method according to the first embodiment is implemented when the processor 22 of the computer device 20 executes a computer program in this embodiment.

Example four

The present embodiment also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, a server, an App application mall, etc., on which a computer program is stored, which when executed by a processor implements corresponding functions. The computer-readable storage medium of this embodiment stores the image processing apparatus 10 of the second embodiment, and when executed by a processor, implements the image processing method of the first embodiment.

The computer-readable storage medium of the present embodiment is used in an image processing apparatus, and when executed by a processor, implements the image processing method of the first embodiment.

Further, the computer-readable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the image data (e.g., the preset image) in the image processing method may be stored in the blockchain node.

The block chain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The invention is operational with numerous general purpose or special purpose computing system environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like. The invention may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.

Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner.

The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims

1. An image processing method, comprising:

acquiring an image to be processed;

2. The image processing method according to claim 1, wherein the processing the image to be processed by using the residual encoder to obtain residual image information comprises:

3. The image processing method according to claim 1, wherein said converting the image to be processed into encoded image information by using an image encoder comprises:

4. The method according to claim 1, wherein before performing the decoding by the image decoder according to the residual image information and the encoded image information to obtain the target image, the method further comprises:

5. The image processing method according to claim 1, wherein said decoding with an image decoder according to the residual image information and the encoded image information to obtain a target image comprises:

6. The image processing method according to claim 1, further comprising, after the acquiring the image to be processed:

7. An image processing apparatus characterized by comprising:

the acquisition module is used for acquiring an image to be processed;

the residual coding module is used for processing the image to be processed by adopting a residual coder to obtain residual image information;

the image coding module is used for converting the image to be processed into coded image information by adopting an image coder;

and the decoding module is used for decoding by adopting an image decoder according to the residual image information and the coded image information to obtain a target image.

8. The apparatus according to claim 7, wherein the residual encoding module obtains a pixel value of the image to be processed, subtracts a pixel value of a preset image from a pixel value of the image to be processed to obtain the residual image data, and encodes the residual image data by using the residual encoder to obtain the residual image information.

9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method of any of claims 1 to 6 are implemented by the processor when executing the computer program.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 6.