WO2021013139A1

WO2021013139A1 - Image processing method and device

Info

Publication number: WO2021013139A1
Application number: PCT/CN2020/103150
Authority: WO
Inventors: 任冬伟; 左旺孟; 秦超; 陈帅
Original assignee: 华为技术有限公司
Priority date: 2019-07-23
Filing date: 2020-07-21
Publication date: 2021-01-28
Also published as: CN110503611A

Abstract

The present application provides an image processing method and device, capable of quickly deblurring a real high-resolution image without distortion and improving the image quality. The image processing method comprises: obtaining a first image, wherein the first image is a high-resolution blurred image; performing down-sampling on the first image to obtain a second image; determining a first blurring kernel according to the second image; and obtaining a third image according to the first blurring kernel, the third image being a deblurred image having the same resolution as the first image.

Description

Image processing method and device

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on July 23, 2019, with the application number of 201910666379.2. The application title is "Image Processing Method and Apparatus", the entire content of which is incorporated into this application by reference.

Background technique

As the resolution of the camera of an electronic device becomes higher and higher, the number of picture pixels generally reaches more than 10M. When the camera of an electronic device is in an unstable state, for example, handshaking or dark light conditions, it is easy to take a blurred picture, which affects the quality of the picture, and brings great inconvenience to the recognition and analysis of the content of the image information.

The traditional deblurring technology can better realize the deblurring of the image and obtain a deblurred image with better quality. However, as the resolution of the picture increases, the amount of calculation based on the traditional deblurring method is very large, the deblurring speed is very slow and time-consuming. With the rise of deep learning technology, deblurring technology based on deep learning has developed rapidly. However, since the training set when training the deblurring model in the deep learning deblurring method is a blurred image synthesized by the algorithm, it is not a real image. When the deblurred image is different from the image in the training set, the training model is used to obtain The deblurring image will have distortion, and the deblurring algorithm based on deep learning takes too much time to deblur high-resolution images, which affects the real-time deblurring.

Therefore, how to quickly deblur real high-resolution images without distortion and improve image quality has become an urgent problem to be solved.

Summary of the invention

This application provides an image processing method and device, which can quickly deblur real high-resolution images without distortion and improve image quality.

In a first aspect, an image processing method is provided, including: acquiring a first image, which is a high-resolution blurred image; down-sampling the first image to obtain a second image; and according to the second image Determine a first blur kernel; obtain a third image according to the first blur kernel, where the third image is a deblurred image with the same resolution as the first image.

In the above technical solution, the obtained high-resolution blurred image is down-sampled to obtain a low-resolution blurred image, the first blur kernel is estimated based on the obtained low-resolution blurred image, and the obtained first blur kernel is compared with the original obtained High-resolution blurred image Deblurred images with the same resolution, thereby achieving improved image quality while reducing the time-consuming deblurring.

With reference to the first aspect, in some possible implementations of the first aspect, determining the first blur kernel according to the second image includes: acquiring an edge map of the second image; determining at least one first region of the edge map , Wherein each first region in the at least one first region is an edge salient region of the edge map; the first blur kernel is determined according to the at least one first region.

First, the edge detection algorithm is used to obtain the edge image of the second image, and then one or more regions containing the salient edge are intercepted to estimate the first blur kernel. For example, the blur kernel of the second image is estimated according to the iterative adaptive prior model of the gradient domain as shown in the following formula:

among them,

To clear the gradient map of the image X, the parameters p and λ are fuzzy and estimated iterative optimization parameters. Through iterative optimization, the first fuzzy kernel k can be obtained.

In some possible implementations, in order to obtain a more ideal restoration effect, an anisotropic total variation (TV) model can be used to make the estimated first blur kernel more accurate. Thereby, a more accurate first blur kernel is obtained.

It should be noted that in the foregoing process of estimating the first blur kernel, when acquiring the edge of the second image, any existing edge detection algorithm, such as Sobel algorithm, Canny algorithm, Laplacian algorithm, etc., can be used. The embodiments of the present application There are no restrictions on the specific algorithm, as long as a more accurate edge image can be obtained.

In addition, when estimating the first blur kernel, the above-mentioned formula is only an exemplary description, and the embodiment of the present application does not specifically limit the method for estimating the first blur kernel.

With reference to the first aspect, in some possible implementations of the first aspect, obtaining a third image according to the first blur kernel includes: performing non-blind deblurring on the second image according to the first blur kernel to obtain a fourth image ; Up-sampling the fourth image to obtain the third image.

After the first blur kernel is estimated, the second image is deblurred according to the first blur kernel to obtain a fourth image, where the fourth image is a deblurred image with the same resolution as the second image.

In some possible implementation manners, even after a series of subsequent processing, the estimated first fuzzy kernel inevitably still has a certain error. In order to reduce the ringing effect caused by the estimation error of the first blur kernel, a bilateral filter can be used to process the deblurred fourth image, and the response value of the bilateral filter can be subtracted, so that the ringing effect can be effectively suppressed. The influence of the estimation error of the first blur kernel on the restoration result is reduced, thereby obtaining a more ideal clear image.

After deblurring the low-resolution blurred image (second image) according to the estimated first blur check, a deblurred image (fourth image) with the same resolution as the second image is obtained. In order to obtain an image with the same resolution as the originally acquired high-resolution blurred image (the first image), the obtained fourth image needs to be up-sampled corresponding to the multiple of the down-sampling, so as to obtain the same resolution as the first image The image after deblurring (third image).

Among them, "upsampling the obtained fourth image corresponding to the multiple of the downsampling" should be understood as: if the second image obtained by downsampling the first image is 1/N of the size of the first image, then Four images should be interpolated by N times to obtain the third image, where N can be any positive integer.

In addition, the embodiment of the present application does not make any limitation on the specific interpolation process of upsampling, and any interpolation algorithm in existing algorithms can be used, such as nearest neighbor interpolation, bilinear quadratic interpolation, bicubic interpolation, etc.

In the above technical solution, the acquired high-resolution blurred image is down-sampled to obtain a low-resolution blurred image, the first blur kernel is estimated according to the obtained low-resolution blurred image, and the down-sampled low resolution is calculated according to the estimated first blur kernel. Deblurring the high-resolution blurred image, and finally up-sampling the obtained low-resolution deblurred image to obtain a deblurred image with the same resolution as the initial blurred image, thereby achieving rapid deblurring of the high-resolution blurred image. The blur kernel estimation method reduces the distortion of the result of the deblurring method based on deep learning, and improves the image quality while reducing the time-consuming deblurring.

With reference to the first aspect, in some possible implementations of the first aspect, the obtaining a third image according to the first blur kernel further includes: up-sampling the first blur kernel to obtain a second blur kernel; The second blur kernel performs non-blind deblurring on the first image to obtain the third image.

In order to further reduce the time-consuming deblurring, the first blur kernel can be up-sampled to obtain the second blur kernel. Since the first blur kernel is relatively small compared to the deblurred image, the algorithm complexity of upsampling the first blur kernel is much lower than the algorithm complexity of upsampling the fourth image to obtain the third image, which greatly simplifies the upper The sampling process reduces the time-consuming process of deblurring.

It should be understood that in order to finally obtain a deblurred image with the same resolution as the first image (that is, the third image), up-sampling corresponding to the multiple of the down-sampling should be performed.

Among them, "upsampling corresponding to the multiple of downsampling" should be understood as: if the second image obtained by downsampling the first image is 1/N of the size of the first image, then the first blur kernel should be performed N times the interpolation to obtain the second fuzzy kernel, where N can be any positive integer.

In the above technical solution, the obtained high-resolution blurred image is down-sampled to obtain a low-resolution blurred image, the first blur kernel is estimated based on the obtained low-resolution blurred image, and the first blur kernel is up-sampled to obtain the second blur According to the obtained second blur check, the originally acquired high-resolution blurred image is deblurred to obtain a deblurred image with the same resolution as the original blurred image, thereby achieving rapid deblurring of the high-resolution blurred image. Since the first blur kernel is smaller than the fourth image, directly up-sampling the second blur kernel can reduce the complexity of the up-sampling process, thereby reducing the time-consuming deblurring. In addition, by directly performing deblurring processing on the high-resolution blurred image, the image distortion caused by upsampling can be reduced, so that the image quality can be improved while reducing the time-consuming deblurring.

With reference to the first aspect, in some possible implementations of the first aspect, determining the first blur kernel based on the second image includes: performing denoising processing on the second image to obtain a fifth image, which is compared with The second image has the same resolution; the first blur kernel is determined according to the fifth image.

By denoising the down-sampled second image, the noise interference caused by the camera is reduced, and the image distortion is reduced, and then the first blur kernel is estimated based on the denoised fifth image to reduce the first blur The error of the kernel estimation is thereby reduced based on the distortion of the deblurred image obtained by the first blur kernel.

With reference to the first aspect, in some possible implementations of the first aspect, the method further includes: performing denoising processing on the third image to obtain a sixth image, the sixth image having the same resolution as the third image.

Optionally, in some possible implementation manners, in order to avoid noise interference to the deblurring result, denoising processing may be performed on the obtained third image, thereby reducing image distortion.

In a second aspect, an image processing apparatus is provided, and the apparatus is configured to execute the foregoing first aspect or the method in any possible implementation of the first aspect. Specifically, the device may include a module for executing the method in the first aspect or any possible implementation of the first aspect.

In a third aspect, an image processing apparatus is provided. The apparatus includes a memory and a processor, the memory is used to store instructions, and the processor is used to execute instructions stored in the memory and perform processing on the images stored in the memory. Execution of the instructions enables the processor to execute the first aspect or the method in any possible implementation manner of the first aspect.

In a fourth aspect, a computer-readable storage medium is provided, and instructions are stored in the computer-readable storage medium, and when the instructions are run on a computer, the computer executes the first aspect or any possible aspect of the first aspect. The method in the implementation mode.

In a fifth aspect, a computer program product containing instructions is provided. When the computer program product runs on a computer, the computer executes the method in the first aspect or any possible implementation of the first aspect.

Description of the drawings

Fig. 1 shows a schematic flowchart of an image processing method provided by an embodiment of the present application;

FIG. 2 shows a schematic flowchart of another image processing method provided by an embodiment of the present application;

FIG. 3 shows the processing result of an image processing method provided by an embodiment of the present application;

FIG. 4 shows a schematic flowchart of yet another image processing method provided by an embodiment of the present application;

FIG. 5 shows the processing result of another image processing method provided by an embodiment of the present application;

Fig. 6 shows a schematic block diagram of an image processing apparatus provided by an embodiment of the present application;

Fig. 7 shows a schematic block diagram of another image processing apparatus provided by an embodiment of the present application.

Detailed ways

The technical solution in this application will be described below in conjunction with the drawings.

FIG. 1 shows a schematic flowchart of an image processing method 100 provided by an embodiment of the present application. It should be understood that FIG. 1 shows the steps or operations of the image processing method, but these steps or operations are only examples, and the embodiment of the present application may also perform other operations or variations of each operation in FIG. 1. In addition, the various steps in FIG. 1 may be performed in a different order from that presented in FIG. 1, and it is possible that not all operations in FIG. 1 are to be performed.

S110: Acquire a first image.

The first image is a high-resolution blurred image. For example, the first image is obtained by shooting with an electronic device, where the electronic device may be a smart phone, a camera, a tablet computer, etc. with a high-resolution camera, and the embodiment of the present application does not limit the method of obtaining the first image.

S120, down-sampling the first image to obtain a second image.

Since the acquired first image is a high-resolution blurred image and the image is relatively large, if the first image is directly de-blurred, the amount of calculation will be too large and the de-blurring will take a long time. Therefore, in order to reduce the complexity of the deblurring process and reduce the time consumption, the first image may be down-sampled to obtain the second image, and the second image may be deblurred.

Wherein, when down-sampling the first image, the first image can be down-sampled to 1/2 of the first image, that is, the pixel points of the obtained second image are half of the pixel points of the first image. Alternatively, the first image may be down-sampled to 1/3 of the first image, that is, the pixels of the second image obtained are 1/3 of the pixels of the first image, and the specific sampling multiple is that in this embodiment of the application. It is not limited.

In some optional implementations, in order to reduce the noise interference of the camera, after the second image is obtained, the second image can be denoised first to obtain a low-resolution blurred image with less interference (ie, the fifth image ). For example, it may be based on a denoising model of a convolutional neural network (CNN), or a denoising model in a traditional method, which is not limited in the embodiment of the present application.

In other optional implementation manners, it is also possible to directly perform denoising processing on the acquired first image, thereby reducing noise interference in the first image.

S130: Determine a first blur kernel according to the second image.

By down-sampling the high-resolution blurred image, that is, the first image, the low-resolution second image is obtained. In order to reduce the time-consuming deblurring process, the second image can be deblurred.

In the prior art, according to whether the blur kernel is known, image deblurring is divided into blind image deblurring (BID) and non-blind image deblurring (NBID). Among them, blind image deblurring means that only the image itself is blurred when the blur kernel is unknown. But non-blind image deblurring means that the blur kernel is known and only the process of image restoration is required. In the embodiments of the present application, in order to obtain a clear image after deblurring without distortion, a non-blind image deblurring method is adopted.

First, estimate the first blur kernel according to the image that needs to be deblurred, that is, the second image. First, the edge detection algorithm is used to obtain the edge image of the second image, and then one or more regions containing the salient edge are intercepted to estimate the first blur kernel. For example, according to the iterative adaptive prior model of the gradient domain shown in formula (1), the blur kernel of the second image is estimated:

In formula (1),

In addition, when estimating the first fuzzy kernel, the formula shown in formula (1) is only an exemplary description, and the embodiment of the present application does not specifically limit the method for estimating the first fuzzy kernel.

S140: Obtain a third image according to the first blur kernel.

After the first blur kernel is estimated, a third image can be obtained based on the first blur kernel, where the third image is a deblurred image with the same resolution as the first image.

FIG. 2 shows a schematic flowchart of an image processing method 200 provided by an embodiment of the present application. It should be understood that FIG. 2 shows the steps or operations of the image processing method, but these steps or operations are only examples, and the embodiment of the present application may also perform other operations or variations of each operation in FIG. 2. In addition, the various steps in FIG. 2 may be performed in a different order from that presented in FIG. 2, and it is possible that not all operations in FIG. 2 are to be performed.

S210: Acquire a first image.

S220: Down-sampling the first image to obtain a second image.

S230: Determine a first blur kernel according to the second image.

Among them, step S210 to step S230 are the same as step S110 to step S130. For detailed description, please refer to step S110 to step S130, which will not be repeated here for brevity.

Wherein, the third image is a deblurred image with the same resolution as the first image.

In the above technical solution, a low-resolution blurred image is obtained by down-sampling a high-resolution blurred image, and the first blur kernel is estimated based on the low-resolution blurred image, and deblurring is performed according to the estimated first blur kernel. Through processing, a deblurred image with the same resolution as the originally acquired high-resolution blurred image is obtained, thereby achieving improved image quality while reducing the time-consuming deblurring.

S240: Deblur the second image according to the estimated first blur check to obtain a fourth image.

After step S230, after the first blur kernel is estimated, the second image is deblurred according to the first blur check to obtain a fourth image, where the fourth image is a deblurred image with the same resolution as the second image.

S250: Up-sampling the fourth image to obtain a third image.

After deblurring the low-resolution blurred image (second image) according to the estimated first blur check, a deblurred image (fourth image) with the same resolution as the second image is obtained. In order to obtain an image with the same resolution as the originally acquired high-resolution blurred image (first image), it is necessary to perform up-sampling on the obtained fourth image corresponding to the multiple of the down-sampling in step S220, so as to obtain a resolution from the first image. The deblurred image with the same rate (third image).

Wherein, "up-sampling the obtained fourth image corresponding to the multiple of the down-sampling in step S220" should be understood as: if the second image obtained by down-sampling the first image in step S220 is 1/N of the first image If the size is small, in S240, the fourth image should be interpolated by N times to obtain the third image, where N can be any positive integer.

In some possible implementation manners, the method 200 may further include S260, performing denoising processing on the third image to obtain the sixth image.

In order to reduce noise interference caused by upsampling and make the obtained third image more accurate, the third image may be denoised to obtain a sixth image with the same resolution as the third image. For example, it may be based on a denoising model of a convolutional neural network (CNN), or a denoising model in a traditional method, which is not limited in the embodiment of the present application.

FIG. 3 shows the result of the method 200 processing. As shown in Figure 3, Figure 3(a) is the acquired high-resolution blurred image (ie the first image), and Figure 3(b) is the low-resolution blurred image (ie the first image) after downsampling the high-resolution blurred image. Two images), where the number of pixels in the second image is 1/2 of the number of pixels in the first image. Fig. 3(c) is the first blur kernel estimated according to the second image, and Fig. 3(d) is the deblurred image with the same resolution as the second image obtained after deblurring the second image according to the first blur check. Four images), Figure 3(e) is a deblurred image (that is, the third image) obtained after up-sampling Figure 3(d) with the same resolution as the originally acquired first image. It can be seen from Fig. 3 that after deblurring, the image quality of Fig. 3(e) is improved compared to Fig. 3(a).

FIG. 4 shows another image processing method 300 provided by an embodiment of the present application. It should be understood that FIG. 4 shows the steps or operations of the image processing method, but these steps or operations are only examples, and the embodiment of the present application may also perform other operations or variations of each operation in FIG. 4. In addition, the various steps in FIG. 4 may be performed in a different order from that presented in FIG. 4, and it is possible that not all operations in FIG. 4 are to be performed.

S310: Acquire a first image.

S320: Down-sampling the first image to obtain a second image.

S330: Determine a first blur kernel according to the second image.

Among them, step S310 to step S330 are the same as step S110 to step S130. For details, please refer to the description of S110 to S130, which will not be repeated here for brevity.

S340: Up-sampling the first blur kernel to obtain a second blur kernel.

After step S330, the first blur kernel is obtained. In order to further reduce the time-consuming deblurring, the first blur kernel can be up-sampled to obtain the second blur kernel. Since the first blur kernel is relatively small compared to the deblurred image, the algorithm complexity of upsampling the first blur kernel is much lower than the algorithm complexity of the third image obtained by upsampling the fourth image in S150, which is greatly simplified The up-sampling process reduces the time-consuming process of the entire de-blurring process.

It should be understood that, in order to finally obtain a deblurred image with the same resolution as the first image (that is, the third image), up-sampling corresponding to the multiple of down-sampling in step S320 should be performed in step S340.

Among them, "step S340 should perform up-sampling corresponding to the multiple of down-sampling in step S320" should be understood as: if the second image obtained by down-sampling the first image in step S320 is 1/N of the size of the first image , Then in S340, the first fuzzy kernel should be interpolated by N times to obtain the second fuzzy kernel, where N can be any positive integer.

S350: Deblur the first image according to the second blur check to obtain a third image.

After step S340, after the second blur kernel is obtained by up-sampling the first blur kernel, the first image is deblurred according to the second blur kernel to obtain a third image, where the third image is the same as the first acquired first image. A deblurred image with the same image resolution.

In some possible implementations, even after a series of subsequent processing, the estimated first blur kernel inevitably still has a certain error. When the first blur kernel is up-sampled to obtain the second blur kernel, it will also Introduce a part of the error. In order to reduce the ringing effect caused by the estimation error of the second blur kernel, a bilateral filter can be used to process the deblurred third image and subtract the response value of the bilateral filter, which can effectively suppress The ringing effect reduces the influence of the estimation error of the second blur kernel on the restoration result, thereby obtaining a more ideal clear image.

In some possible implementation manners, the method 300 may further include S360, performing denoising processing on the third image to obtain the sixth image.

FIG. 5 shows the result of the method 300 processing. As shown in Figure 5, Figure 5(a) is the acquired high-resolution blurred image (i.e. the first image), and Figure 5(b) is the low-resolution blurred image (i.e. the first image) after downsampling the high-resolution blurred image. Two images), where the number of pixels in the second image is 1/2 of the number of pixels in the first image. Fig. 5(c) is the first blur kernel estimated from the second image, Fig. 5(d) is the second blur kernel obtained after upsampling the first blur kernel, and Fig. 5(e) is the check according to the second blur Fig. 5(a) The deblurred image after deblurring (ie the third image). It can be seen from Figure 5 that after deblurring, the image quality of Figure 5(e) is improved compared to Figure 5(a).

The image processing method according to the embodiment of the present application is described in detail above with reference to FIGS. 1 to 5, and the image processing apparatus according to the embodiment of the present application will be described in detail below in conjunction with FIG. 6 to FIG. 7.

FIG. 6 shows a schematic block diagram of an image processing apparatus 500 provided by an embodiment of the present application. The image processing apparatus includes an acquisition unit 510 and a processing unit 520.

The acquiring unit 510 is configured to acquire a first image, where the first image is a high-resolution blurred image.

The processing unit 520 is configured to down-sample the first image to obtain the second image.

The processing unit 520 is further configured to determine the first blur kernel according to the second image.

The processing unit 520 is further configured to obtain a third image according to the first blur kernel, where the third image is a deblurred image with the same resolution as the first image.

Optionally, the acquiring unit 510 is specifically configured to acquire an edge map of the second image.

Optionally, the processing unit 520 is specifically configured to: determine at least one first region of the edge map, wherein each first region in the at least one first region is a prominent edge region of the edge map.

Optionally, the processing unit 520 is specifically configured to: determine the first blur kernel according to the at least one first region.

Optionally, the processing unit 520 is further specifically configured to:

Perform non-blind deblurring on the second image according to the first blur kernel to obtain a fourth image;

Up-sampling the fourth image to obtain the third image.

Optionally, the processing unit 520 is further specifically configured to:

Up-sampling the first blur kernel to obtain a second blur kernel;

The third image is obtained by unblindly deblurring the first image according to the second blur kernel.

Optionally, the processing unit 520 is further specifically configured to:

Performing denoising processing on the second image to obtain a fifth image with the same resolution as the second image;

The first blur kernel is determined according to the fifth image.

Optionally, the processing unit 520 is further specifically configured to: perform denoising processing on the third image to obtain a sixth image, where the sixth image has the same resolution as the third image.

It should be understood that the image processing apparatus 500 provided in the embodiment of the present application is used to implement any method in the foregoing method embodiments. For specific details, refer to the foregoing method, and details are not described herein again.

It should be noted that in this embodiment of the present application, the acquiring unit 510 may be implemented by a communication interface, and the processing unit 520 may be implemented by a processor. As shown in FIG. 7, the image processing apparatus 600 may include a processor 610, a memory 620, and a communication interface 630. The memory 620 may be used to store codes executed by the processor 610, and the processor 610 may be used to process data or programs.

In the implementation process, the steps of the foregoing method may be completed by an integrated logic circuit of hardware in the processor 610 or instructions in the form of software. The steps of the method disclosed in the embodiments of the present invention may be directly embodied as being executed and completed by a hardware processor, or executed and completed by a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers. The storage medium is located in the memory 620, and the processor 610 reads the information in the memory 620, and completes the steps of the foregoing method in combination with its hardware. In order to avoid repetition, it will not be described in detail here.

The device 500 shown in FIG. 6 or the device 600 shown in FIG. 7 can implement each process of the image processing method corresponding to the foregoing method embodiment. Specifically, the device 500 or the device 600 can refer to the above description, in order to avoid repetition , I won’t repeat it here.

The embodiment of the present application also provides a computer-readable medium for storing a computer program, and the computer program includes instructions for executing the corresponding method in the foregoing method embodiment.

The embodiments of the present application also provide a computer program product, the computer program product comprising: computer program code, so that the image processing apparatus executes the corresponding method in any of the foregoing method embodiments.

The various embodiments in this application can be used independently or in combination, which is not limited here.

It should be understood that the first, second, etc. descriptions appearing in the embodiments of this application are only for illustration and to distinguish the description objects, and there is no order, nor does it mean that the number of devices in the embodiments of this application is particularly limited. It constitutes any limitation to the embodiments of the present application.

It should also be understood that in the various embodiments of the present application, the size of the sequence number of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not be implemented in this application. The implementation process of the example constitutes any limitation.

A person of ordinary skill in the art may be aware that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.

Those skilled in the art can clearly understand that, for the convenience and conciseness of description, the specific working process of the above-described system, device, and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

The foregoing embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above-mentioned embodiments may be implemented in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on the computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, computer, server, or data center through a (Such as infrared, wireless, microwave, etc.) to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more sets of available media. The usable medium may be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium may be a solid state drive.

A person of ordinary skill in the art may be aware that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed in this document can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.

In the several embodiments provided in this application, it should be understood that the disclosed system, device, and method may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components can be combined or It can be integrated into another system, or some features can be ignored or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

In addition, the functional units in each embodiment of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.

If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solution of this application essentially or the part that contributes to the existing technology or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .

The above are only specific implementations of this application, but the protection scope of this application is not limited to this. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application. Should be covered within the scope of protection of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

An image processing method, characterized in that it comprises:

Acquiring a first image, the first image being a high-resolution blurred image;

Down-sampling the first image to obtain a second image;

Determining a first blur kernel according to the second image;

A third image is obtained according to the first blur kernel, where the third image is a deblurred image with the same resolution as the first image.
The method according to claim 1, wherein the determining a first blur kernel according to the second image comprises:

Acquiring an edge map of the second image;

Determining at least one first region of the edge map, wherein each first region in the at least one first region is a prominent edge region of the edge map;

The first blur kernel is determined according to the at least one first region.
The method according to claim 1 or 2, wherein the obtaining a third image according to the first blur kernel comprises:

Perform non-blind deblurring on the second image according to the first blur check to obtain a fourth image;

Up-sampling the fourth image to obtain the third image.
The method of claim 1 or 2, wherein the obtaining a third image according to the first blur kernel further comprises:

Up-sampling the first blur kernel to obtain a second blur kernel;

Perform non-blind deblurring of the first image according to the second blur check to obtain the third image.
The method according to any one of claims 1 to 4, wherein the determining a first blur kernel according to the second image comprises:

Performing denoising processing on the second image to obtain a fifth image, where the fifth image has the same resolution as the second image;

The first blur kernel is determined according to the fifth image.
The method according to any one of claims 1 to 5, wherein the method further comprises:

Performing denoising processing on the third image to obtain a sixth image, the sixth image having the same resolution as the third image.
An image processing device, characterized in that it comprises:

An acquiring unit, configured to acquire a first image, the first image being a high-resolution blurred image;

A processing unit, configured to down-sample the first image to obtain a second image;

The processing unit is further configured to determine a first blur kernel according to the second image;

The processing unit is further configured to obtain a third image according to the first blur kernel, where the third image is a deblurred image with the same resolution as the first image.
The device according to claim 7, wherein the acquiring unit is specifically configured to:

Acquiring an edge map of the second image;

The processing unit is specifically configured to: determine at least one first region of the edge map, wherein each first region in the at least one first region is a prominent edge region of the edge map;

The processing unit is further specifically configured to determine the first blur kernel according to the at least one first region.
The device according to claim 7 or 8, wherein the processing unit is further specifically configured to:

Perform non-blind deblurring on the second image according to the first blur check to obtain a fourth image;

Up-sampling the fourth image to obtain the third image.
The device according to claim 7 or 8, wherein the processing unit is further specifically configured to:

Up-sampling the first blur kernel to obtain a second blur kernel;

Perform non-blind deblurring of the first image according to the second blur check to obtain the third image.
The device according to any one of claims 7 to 10, wherein the processing unit is further specifically configured to:

Performing denoising processing on the second image to obtain a fifth image, where the fifth image has the same resolution as the second image;

The first blur kernel is determined according to the fifth image.
The device according to any one of claims 7 to 11, wherein the processing unit is further specifically configured to:

Performing denoising processing on the third image to obtain a sixth image, the sixth image having the same resolution as the third image.
A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, and when the computer program runs on a processor, the processor executes any one of claims 1 to 6 The steps in the method described in item.
A computer program product, wherein the computer program product includes instructions for executing the method according to any one of claims 1 to 6.