Image optimization method, apparatus, device and storage medium
The disclosure provides an image optimization method, system, and storage medium. The image optimization method includes extracting texture quality information from an input image. The texture quality information indicates a spatial distribution of texture quality in the input image. The image optimization method also includes performing, according to the texture quality information, texture restoration on a set region in the input image to generate a texture restored image.
1. A method for operating an electronic device, the method comprising:
acquiring an input image;
extracting texture quality information from the input image, the texture quality information indicating a spatial distribution of texture quality in the input image, wherein the texture quality information indicating the spatial distribution of texture quality is embodied in the input image as a texture quality map or a probability map of a degree of texture quality of each pixel forming the input image;
based on the texture quality information, setting a region to be restored in the input image; and
performing texture restoration on the set region in the input image, using image information inside the set region and image information outside the set region,
wherein texture quality within the set region is lower than a preset threshold value, and
wherein the image information inside the set region used for the texture restoration comprises texture quality information and a color.
2. The method of claim 1 , wherein extracting the texture quality information from the input image comprises:
performing feature extraction on the input image through a first convolutional neural network (CNN) to obtain the texture quality information.
3. The method of claim 1 , wherein the texture quality information includes a value between 0 and 1.
4. The method of claim 1 , wherein the texture quality information of a boundary position of the set region is smoothed when the texture quality information is binarization information.
5. The method of claim 1 , wherein performing the texture restoration comprises:
performing the texture restoration on the set region through a second CNN based on the texture quality information.
6. The method of claim 1 , further comprising performing texture feature enhancement on the restored texture of the input image.
7. The method of claim 6 , wherein performing the texture feature enhancement comprises performing the texture feature enhancement on the restored texture of the input image via a residual network using the restored texture of the input image and a residual output by the residual network.
8. The method of claim 7 , wherein the residual network comprises concatenated convolutional units.
9. The method of claim 8 , wherein any one of the concatenated convolutional units of the residual network comprises a plurality of concatenated dilated convolutional layers.
10. The method of claim 9 , wherein at least two dilated convolutional layers of the plurality of concatenated dilated convolutional layers include different dilated ratios.
11. An electronic device comprising:
a first processor; and
a memory connected to the first processor configured to store machine-readable instructions that when executed by the first processor to cause the first processor to:
acquire an input image,
extract texture quality information from the input image, the texture quality information indicating a spatial distribution of texture quality in the input image, wherein the texture quality information indicating the spatial distribution of texture quality is embodied in the input image as a texture quality map or a probability map of a degree of texture quality of each pixel forming the input image,
based on the texture quality information, set a region to be restored in the input image, and
perform texture restoration on the set region in the input image, using image information inside the set region and image information outside the set region,
wherein texture quality within the set region is lower than a preset threshold value, and
wherein the image information inside the set region used for the texture restoration comprises texture quality information and a color.
12. The electronic device of claim 11 , further comprising a second processor configured to process an image,
wherein the instructions that when executed by the first processor further cause the first processor to acquire the input image from the second processor.
13. A non-transitory machine-readable storage medium comprising machine-readable instructions that when executed by a processor to cause the processor to:
acquire an input image,
extract texture quality information from the input image, the texture quality information indicating a spatial distribution of texture quality in the input image, wherein the texture quality information indicating the spatial distribution of texture quality is embodied in the input image as a texture quality map or a probability map of a degree of texture quality of each pixel forming the input image,
based on the texture quality information, set a region to be restored in the input image, and
perform texture restoration on the set region in the input image, using image information inside the set region and image information outside the set region,
wherein texture quality within the set region is lower than a preset threshold value, and
wherein the image information inside the set region used for the texture restoration comprises texture quality information and a color.