WO2020107651A1

WO2020107651A1 - Image processing method

Info

Publication number: WO2020107651A1
Application number: PCT/CN2019/070511
Authority: WO
Inventors: 赖庆鸿
Original assignee: 深圳市华星光电半导体显示技术有限公司
Priority date: 2018-11-29
Filing date: 2019-01-04
Publication date: 2020-06-04
Also published as: CN109658364A

Abstract

Provided in the present invention is an image processing method. The image processing method of the present invention comprises: acquiring a full-image degree of interest weight map of an original image; dividing the original image into a plurality of original blocks; using the full-image degree of interest weight map and a preset greyscale conversion formula to respectively perform histogram equalisation processing on the plurality of original blocks to form a plurality of converted blocks; and performing splicing processing on the plurality of converted blocks to form a processed image; the contrast ratio of the image can be dynamically enhanced according to the degree of human visual interest, increasing the image quality.

Description

Specification Title: Image processing method

Technical field

[0001] The present invention relates to the field of display technology, and in particular, to an image processing method.

Background technique

[0002] Image enhancement technology is a type of image processing technology, which can significantly improve image quality, make the image content more layered and subjective observation effect more in line with people's needs. In real life, the original image often has various problems, for example: the aperture is too small when taking a picture, resulting in a dark image; the contrast of the scene is low, so that the focus of the image is not prominent; overexposure leads to abnormal images, and the photo is white. The image enhancement technology can effectively solve the above problems and improve the display quality.

[0003] Common image enhancement technologies include: saturation enhancement and contrast enhancement. Compared with saturation enhancement, contrast enhancement receives more attention. Contrast enhancement is to adjust the gray scale distribution of the image and increase the distribution range of the gray scale of the image to improve the contrast of the whole or part of the image and improve the visual effect.

[0004] In common contrast enhancement algorithms, histogram equalization is often used to process images to enhance contrast. The basic idea is to transform the original image histogram into a uniformly distributed form, that is, the number of compressed pixels is relatively high. Less gray scale and expand the gray scale with more pixels, this can increase the dynamic range of the pixel gray value, so as to achieve the effect of enhancing the overall image contrast.

[0005] When performing contrast enhancement processing on an image by using a histogram equalization method, it generally includes two forms of a full image form and a block division form. The full image format refers to directly performing histogram equalization on the entire image. Since the output image is affected by the original image, when there are a large number of similar pixels in the original image (such as large bright areas, dark areas, sky scenes, etc.), it will take up more The large range of gray values results in insufficient contrast enhancement of image details. The block division method refers to dividing the original input image into blocks, and performing local histogram equalization processing on each block, which can solve the image details brought by the full-histogram histogram equalization to a certain extent. The problem of contrast enhancement is adequate, but the image contrast enhancement cannot be performed according to the degree of human visual attention, and the effect of enhancing contrast is still very limited. Summary of the invention

technical problem [0006] An object of the present invention is to provide an image processing method that can significantly improve the contrast of an image and improve the quality of the image.

Solution to the problem

Technical solution

[0007] To achieve the above object, the present invention provides an image processing method, including the following steps:

[0008] Step S1, providing an original image; the original image includes a plurality of pixels, each pixel has a corresponding original gray scale;

[0009] Step S2: Obtain a full map attention degree weight map of the original image; the full map attention degree weight map includes mapping points respectively corresponding to a plurality of pixels, and each mapping point has a corresponding weight value;

[0010] Step S3: Divide the original image into multiple original blocks;

[0011] Step S4: Perform histogram equalization processing on multiple original blocks to form multiple conversion blocks, respectively, using a full-image attention weight map and a preset gray-scale conversion formula;

[0012] Step S5: Perform stitching processing on a plurality of conversion blocks to form a processed image.

[0013] In the step S4, the conversion gray scale corresponding to the original gray scale of each pixel in each original block is calculated by using the full map attention weight map and the preset gray scale conversion formula respectively and using the conversion gray scale The original gray scale is replaced to form multiple conversion blocks.

[0014] The preset gray scale conversion formula is:

[]

[0015] where,

Is the converted gray scale of pixels with the original gray scale of k in the original block,

%

Is the sum of the gray levels of all pixels in the original block with the original gray level j after being multiplied by the corresponding weight values, n is the gray level of all pixels in the original block after being multiplied by the corresponding weight values, respectively Summed values [0016] k is a positive integer, and the value range of k is 0~L-1, where L is the total number of available gray levels of the original image.

[0017] The image processing method further includes step S5, performing stitching processing on a plurality of conversion blocks to form a processed image.

[0018] In the step S5, a plurality of conversion blocks are spliced by linear interpolation.

[0019] In the step S3, the original image is divided into a plurality of original blocks arranged in an array.

[0020] In the step S5, the plurality of conversion blocks are divided into four sub-blocks arranged in two rows and two columns, so that the sub-blocks of the plurality of conversion blocks are arranged in an array, through linear interpolation The specific method of splicing multiple conversion blocks is as follows: the sub-blocks in the first row and first column, the sub-blocks in the first row and last column, the sub-blocks in the last row and first column, and the last line in the last row The sub-blocks in one column are not linearly interpolated. For the sub-blocks in the first row, first column, last row, and last column, except the sub-blocks in the first row and first column, and the sub-regions in the first row and last column Blocks, sub-blocks in the last row and first column, and sub-blocks other than the sub-blocks in the last row and last column are subjected to a single linear interpolation process, except for the first row, first column, and last row in the sub-block array , The sub-blocks other than the sub-blocks in the last column are bilinearly interpolated.

[0021] The single linear interpolation processing is horizontal single linear interpolation processing.

[0022] The single linear interpolation process is a horizontal vertical single linear interpolation process.

[0023] In the step S2, the GBVS algorithm is used to obtain the full image attention weight map of the original image.

[0024] The present invention also provides an image processing method, including the following steps:

[0025] Step S1, providing an original image; the original image includes a plurality of pixels, each pixel having a corresponding original gray scale;

[0026] Step S2: Obtain a full map attention level weight map of the original image; the full map attention level weight map includes mapping points respectively corresponding to a plurality of pixels, and each mapping point has a corresponding weight value;

[0027] Step S3: Divide the original image into multiple original blocks;

[0028] Step S4: Perform histogram equalization processing on a plurality of original blocks to form a plurality of conversion blocks, respectively, using a full-image attention weight map and a preset gray-scale conversion formula;

[0029] Step S5: Perform stitching processing on multiple conversion blocks to form a processed image;

[0030] In the step S4, the conversion gray scale corresponding to the original gray scale of each pixel in each original block is calculated by using the full map attention weight map and the preset gray scale conversion formula respectively and using the conversion gray scale To the original The gray scale is replaced to form multiple conversion blocks;

[0031] In the step S5, a plurality of conversion blocks are spliced by linear interpolation.

Beneficial effects of invention

Beneficial effect

[0032] Beneficial effects of the present invention: The image processing method of the present invention obtains the full image attention weight map of the original image, and divides the original image into multiple original blocks, using the full image attention weight map and preset The gray-scale conversion formula of each performs histogram equalization processing on multiple original blocks to form multiple conversion blocks, and then performs stitching processing on the multiple conversion blocks to form a processed image, which can be dynamically based on the degree of human visual attention. Improve the contrast of the image and the quality of the image.

Brief description of the drawings

BRIEF DESCRIPTION

[0033] In order to further understand the features and technical content of the present invention, please refer to the following detailed description and drawings of the present invention, however, the drawings are provided for reference and explanation only, and are not intended to limit the present invention.

[0034] In the drawings,

[0035] FIG. 1 is a flowchart of an image processing method of the present invention;

[0036] FIG. 2 is a schematic diagram of step S1 of the image processing method of the present invention;

[0037] FIG. 3 is a schematic diagram of step S2 of the image processing method of the present invention;

[0038] FIG. 4 is a schematic diagram of step S3 of the image processing method of the present invention;

[0039] FIG. 5 is a schematic diagram of step S4 of the image processing method of the present invention;

6 is a schematic diagram of step S5 of the image processing method of the present invention.

Invention Example

Embodiments of the invention

[0041] In order to further illustrate the technical means adopted by the present invention and its effects, the following will be described in detail in conjunction with the preferred embodiments of the present invention and the accompanying drawings.

[0042] Please refer to FIG. 1, the present invention provides an image processing method, including the following steps:

[0043] Step S1, please refer to FIG. 2 to provide an original image 10. The original image 10 includes a plurality of pixels, and each pixel has a corresponding original gray level. [0044] Step S2. Please refer to FIG. 3 to obtain the full image attention weight map 20 of the original image 10. The full map attention degree weight map 20 includes mapping points corresponding to a plurality of pixels, and each mapping point has a corresponding weight value.

[0045] Specifically, in the step S2, an attention selection model (Graph-based Visual Saliency

, GBVS) algorithm to obtain the full image attention weight map 20 of the original image.

[0046] Step S3. Referring to FIG. 4, the original image 10 is divided into a plurality of original blocks 11.

[0047] Specifically, in the embodiment shown in FIG. 4, in the step S3, the original image 10 is divided into a plurality of original blocks 11 arranged in an array. Of course, the original image 10 is divided into the original block 11 can also be used in other ways

[0048] Step S4. Please refer to FIG. 5, using the full-map attention weight map 20 and the preset gray-scale conversion formula to perform histogram equalization processing on a plurality of original blocks 11 respectively to form a plurality of conversion blocks 31.

[0049] Specifically, in the step S4, the conversion gray scale corresponding to the original gray scale of each pixel in each original block 11 is calculated by using the full map attention weight map 20 and the preset gray scale conversion formula respectively And the conversion gray scale is used to replace the original gray scale, thereby forming a plurality of conversion blocks 31.

[0050] Further, the preset gray scale conversion formula is:

[]

[0051] where,

Sk

Is the converted gray scale of the pixels in the original block 11 with the original gray scale of k,

II _I

1 is the sum of the gray levels of all pixels in the original block 11 with the original gray level j and the corresponding weight values, and n is the gray levels of all pixels in the original block 11 and the corresponding weight values The summed value after multiplication. k is a positive integer, and the value of k ranges from 0 to L-1, where L is the total number of available gray levels of the original image 10. [0052] Step S5. Please refer to FIG. 6, performing a stitching process on a plurality of conversion blocks 31 to form a processed image 30.

[0053] Specifically, in the step S5, a plurality of conversion blocks 31 may be stitched by linear interpolation.

[0054] Further, in the embodiment shown in FIG. 6, in the step S5, the plurality of conversion blocks 31 are divided into four sub-blocks 311 arranged in two rows and two columns, so that a plurality of conversion areas The sub-blocks 311 of the block 31 are arranged in an array, and the method of splicing the multiple conversion blocks 31 by linear interpolation is as follows: the sub-block 311 in the first row and the first column, the last in the first row The sub-blocks 311 in one column, the sub-blocks 311 in the last row and first column, and the sub-blocks 311 in the last row and last column do not perform linear interpolation processing, and the sub-regions in the first row, first column, last row, and last column Except for the sub-block 311 in the first row and first column, the sub-block 311 in the first row and last column, the sub-block 311 in the last row and first column, and the sub-block 311 in the last row and last column The sub-block 311 performs a single linear interpolation process, and performs bilinear interpolation on the other sub-blocks 311 in the sub-block 311 array except the sub-blocks 311 in the first row, first column, last row, and last column Insert processing.

[0055] Further, the single linear interpolation processing is horizontal single linear interpolation processing or vertical single linear interpolation processing.

[0056] It should be noted that in the image processing method of the present invention, the original image 10 is provided first, and then the full image attention degree weight map 20 of the original image 10 is obtained, and the full image attention degree weight map 20 includes the original image 10 and the original image. Multiple mapping points corresponding to multiple pixels of 10, each mapping point has a corresponding weight value, and then the original image 10 is divided into a plurality of original blocks 11, using the full map attention degree weight map 20 and the preset The gray scale conversion formula performs histogram equalization processing on multiple original blocks 11 respectively to form multiple conversion blocks 31. Specifically, each original block is calculated by using the full map attention weight map 20 and the preset gray scale conversion formula The conversion gray scale corresponding to the original gray scale of each pixel in the block 11 and using the conversion gray scale to replace the original gray scale, thereby forming a plurality of conversion blocks 31, and finally performing a splicing process on the plurality of conversion blocks 31 to form a process After the image 30, it can effectively avoid the problem of poor contrast enhancement caused by a large range of similar pixels occupying a large gray value range, and at the same time can dynamically improve the contrast of each original block 11 according to the degree of human visual attention, thereby The finally obtained processed image 30 has a better contrast enhancement effect, so that the image has a higher quality.

[0057] In summary, the image processing method of the present invention acquires a full map attention degree weight map of the original image, Divide the original image into multiple original blocks, and use the full image attention weight map and the preset grayscale conversion formula to perform histogram equalization processing on the multiple original blocks respectively to form multiple conversion blocks, and then A plurality of conversion blocks are stitched to form a processed image, which can dynamically improve the contrast of the image and improve the quality of the image according to the degree of human visual attention.

[0058] As mentioned above, those of ordinary skill in the art can make various other corresponding changes and modifications according to the technical solutions and technical concepts of the present invention, and all these changes and modifications should belong to the claims of the present invention. Scope of protection.

Claims

[Claim 1] An image processing method, comprising the following steps:

Step S1: Provide an original image; the original image includes a plurality of pixels, and each pixel has a corresponding original gray scale;

Step S2: Acquire a full image attention weight map of the original image; the full image attention weight map includes mapping points corresponding to multiple pixels, each mapping point having a corresponding weight value;

Step S3: Divide the original image into multiple original blocks;

Step S4: Perform histogram equalization processing on multiple original blocks to form multiple conversion blocks by using the full map attention weight map and the preset gray-scale conversion formula;

Step S5: Perform stitching processing on multiple conversion blocks to form a processed image.

[Claim 2] The image processing method according to claim 1, wherein, in the step S4, each of the original blocks is calculated using the full image attention weight map and the preset grayscale conversion formula The conversion gray scale corresponding to the original gray scale of the pixel and using the conversion gray scale to replace the original gray scale, thereby forming a plurality of conversion blocks.

[Claim 3] The image processing method according to claim 2, wherein the preset gray-scale conversion formula is:

among them,

C

IV

Is the converted gray level of the pixel with the original gray level of k in the original block,

Is the sum of the gray levels of all pixels in the original block with the original gray level j after being multiplied by the corresponding weight values, n is the gray level of all pixels in the original block after being multiplied by the corresponding weight values The summed value.

[Claim 4] The image processing method according to claim 3, wherein k is a positive integer, and the value range of k is 0-L-1, where L is the total number of available gray levels of the original image.

[Claim 5] The image processing method according to claim 1, wherein in the step S5, a plurality of conversion blocks are stitched by linear interpolation.

[Claim 6] The image processing method according to claim 5, wherein in step S3, the original image is divided into a plurality of original blocks arranged in an array.

[Claim 7] The image processing method of claim 6, wherein in the step S5, the plurality of conversion blocks are divided into four sub-blocks arranged in two rows and two columns, so that the plurality of conversion areas The sub-blocks of the block are arranged in an array, and the method of splicing multiple conversion blocks by linear interpolation is as follows: the sub-blocks in the first row and first column, and the sub-areas in the last row and first column Blocks, sub-blocks in the first row and the first column, and sub-blocks in the last row and the last column are not linearly interpolated. Except for the first row, the first row, the first column, the last row, and the last column The sub-blocks in one column, the sub-blocks in the first row and the last column, the sub-blocks in the first row and the first column, and the other sub-blocks other than the sub-blocks in the last row and the last column are subjected to unilinear interpolation processing. Except for the sub-blocks in the first row, the first column, the last row, and the last column in the block array, bilinear interpolation processing is performed.

[Claim 8] The image processing method according to claim 7, wherein the monolinear interpolation processing is horizontal monolinear interpolation processing.

[Claim 9] The image processing method according to claim 7, wherein the monolinear interpolation processing is vertical monolinear interpolation processing.

[Claim 10] The image processing method according to claim 1, wherein in the step S2, GBV is used

The S algorithm obtains the weight map of the attention degree of the whole image of the original image.

[Claim 11] An image processing method, comprising the following steps:

Step S3: Divide the original image into multiple original blocks;

Step S5: Perform stitching processing on multiple conversion blocks to form a processed image; wherein, in step S4, the full image attention weight map and the preset grayscale conversion formula are used to calculate each original block separately The conversion gray scale corresponding to the original gray scale of each pixel of the pixel and using the conversion gray scale to replace the original gray scale to form a plurality of conversion blocks; wherein, in the step S5, the plurality of conversion areas are converted by linear interpolation The blocks are spliced.

[Claim 12] The image processing method of claim 11, wherein the preset gray-scale conversion formula is:

among them,

11 _| Is the sum of the gray levels of all pixels in the original block with the original gray level j after being multiplied by the corresponding weight values, n is the gray level of all pixels in the original block after being multiplied by the corresponding weight values, respectively The summed value.

[Claim 13] The image processing method according to claim 12, wherein k is a positive integer and the value of k ranges from 0 to L-1, where L is the total number of available gray levels of the original image.

[Claim 14] The image processing method according to claim 11, wherein in step S3, the original image is divided into a plurality of original blocks arranged in an array.

[Claim 15] The image processing method of claim 14, wherein in the step S5, the plurality of conversion blocks are divided into four sub-blocks arranged in two rows and two columns, so that the plurality of conversion areas The sub-blocks of the block are arranged in an array, and the method of splicing multiple conversion blocks by linear interpolation is as follows: the sub-blocks in the first row and first column, and the sub-areas in the last row and first column Blocks, sub-blocks in the last row and first column, and sub-blocks in the last row and last column are not linearly interpolated, except for the sub-blocks in the first row, first column, last row, and last column The sub-blocks in one column, the sub-blocks in the first row and the last column, the sub-blocks in the first row and the first column, and the other sub-blocks other than the sub-blocks in the last row and the last column are subjected to unilinear interpolation processing. Except for the sub-blocks in the first row, the first column, the last row, and the last column in the block array, bilinear interpolation processing is performed.