Method and device for image processing, and electronic device

1 . An image processing method comprising:

processing an initial image to obtain a target condition vector, wherein the target condition vector represents degradation feature information of the initial image;

obtaining filter weights in different scale spaces based on the target condition vector;

encoding the initial image to obtain a multi-scale feature map;

processing the multi-scale feature map using the different filter weights to obtain a target residual image; and

performing target processing on the initial image and the target residual image to obtain a target output image.

2 . The image processing method of claim 1 , wherein using different filter weights to process the multi-scale feature map obtained by encoding the initial image to obtain the target residual image includes:

encoding the initial image into a first feature map set, the first feature map set including feature maps of different scales;

converting the first feature map set into a second feature map set by using the corresponding filter weight, the second feature map set including feature maps of different scales; and

sequentially decoding the feature maps in the second feature map set to obtain the target residual image.

3 . The image processing method of claim 2 , wherein converting the first feature map set into the second feature map set by using the corresponding filter weight includes:

performing convolution processing on a first feature map included in the first feature map set by using a first filter weight to obtain a fourth feature map in the second feature map set;

performing convolution processing on a second feature map included in the first feature map set by using a second filter weight to obtain a fifth feature map in the second feature map set; and

performing convolution processing on a third feature map included in the first feature map set by using a third filter weight to obtain a sixth feature map in the second feature map set;

wherein scale spaces corresponding to the first feature map, the second feature map, and the third feature map are different, and scale spaces corresponding to the fourth feature map, the fifth feature map, and the sixth feature map are different.

4 . The image processing method of claim 3 , wherein sequentially decoding the feature maps in the second feature map set to obtain the target residual image includes:

decoding the sixth feature map to obtain a seventh feature map, the scale space corresponding to the seventh feature map being the same as the scale space of the fifth feature map;

accumulating the seventh feature map and the fifth feature map and performing decoding processing to obtain an eighth feature map, the scale space corresponding to the eighth feature map being the same as the scale space of the fourth feature map; and

performing residual processing after accumulating the eighth feature map and the fourth feature map to obtain the target residual image.

5 . The image processing method of claim 1 , wherein obtaining the filter weights in different scale spaces includes:

extracting a third feature map set of the initial image by using the target condition vector and the initial image; and

respectively performing convolution processing and residual processing on each feature map in the third feature map set to obtain the filter weights in different scale spaces.

6 . The image processing method of claim 1 , wherein processing the initial image to obtain the target condition vector includes:

extracting degraded features of the initial image; and

determining the target condition vector based on the degraded features.

7 . The image processing method of claim 5 , wherein extracting the third feature map set of the initial image by using the target condition vector and the initial image includes:

using a first target condition vector to perform convolution and spatial feature conversion processing on the initial image to extract a first scale spatial feature map, the first target condition vector having the same shape as the first scale spatial feature map;

using a second target condition vector to perform convolution and spatial feature conversion processing on the first scale spatial feature map to extract a second scale spatial feature map, the second target condition vector having the same shape as the second scale spatial feature map;

using a third target condition vector to perform convolution and spatial feature conversion processing on the second scale spatial feature map to extract a third scale spatial feature map, the third target condition vector having the same shape as the third scale spatial feature map,

the first target condition vector, the second target condition vector, and the third target condition vector being obtained based on conversion processing of the target condition vector.

8 . The image processing method of claim 5 , wherein extracting the third feature map set of the initial image by using the target condition vector and the initial image includes:

converting the target condition vector into a ninth feature map having the same shape as an input feature map of a different scale space of the initial image;

fusing different ninth feature maps with the input feature map of the corresponding scale space into a tenth feature map;

performing convolution processing on different tenth feature maps to obtain corresponding feature tensors; and

using the feature tensors to perform spatial feature conversion processing on the input feature map corresponding to the scale space to obtain the third feature map set including a multi-scale space feature map.

9 . An image processing device comprising:

a processing module configured to:

process an initial image to obtain a target condition vector, wherein the target condition vector represents degradation feature information of the initial image;

obtain filter weights in different scale spaces based on the target condition vector; and

a determination module configured to:

encode the initial image to obtain a multi-scale feature map;

process the multi-scale feature map using the different filter weights to obtain a target residual image; and

perform target processing on the initial image and the target residual image to obtain a target output image.

10 . The image processing device of claim 9 , wherein the determination module is further configured to:

encode the initial image into a first feature map set, the first feature map set including feature maps of different scales;

convert the first feature map set into a second feature map set by using the corresponding filter weight, the second feature map set including feature maps of different scales; and

sequentially decode the feature maps in the second feature map set to obtain the target residual image.

11 . The image processing device of claim 10 , wherein the determination module is further configured to:

perform convolution processing on a first feature map included in the first feature map set by using a first filter weight to obtain a fourth feature map in the second feature map set;

perform convolution processing on a second feature map included in the first feature map set by using a second filter weight to obtain a fifth feature map in the second feature map set; and

perform convolution processing on a third feature map included in the first feature map set by using a third filter weight to obtain a sixth feature map in the second feature map set;

wherein scale spaces corresponding to the first feature map, the second feature map, and the third feature map are different, and scale spaces corresponding to the fourth feature map, the fifth feature map, and the sixth feature map are different.

12 . The image processing device of claim 11 , wherein the determination module is further configured to:

decode the sixth feature map to obtain a seventh feature map, the scale space corresponding to the seventh feature map being the same as the scale space of the fifth feature map;

accumulate the seventh feature map and the fifth feature map and perform decoding processing to obtain an eighth feature map, the scale space corresponding to the eighth feature map being the same as the scale space of the fourth feature map; and

perform residual processing after accumulating the eighth feature map and the fourth feature map to obtain the target residual image.

13 . The image processing device of claim 9 , wherein the processing module is further configured to:

extract a third feature map set of the initial image by using the target condition vector and the initial image; and

respectively perform convolution processing and residual processing on each feature map in the third feature map set to obtain the filter weights in different scale spaces.

14 . The image processing device of claim 9 , wherein the processing module is further configured to:

extract degraded features of the initial image; and

determine the target condition vector based on the degraded features.

15 . The image processing device of claim 13 , wherein the processing module is further configured to:

use a first target condition vector to perform convolution and spatial feature conversion processing on the initial image to extract a first scale spatial feature map, the first target condition vector having the same shape as the first scale spatial feature map;

use a second target condition vector to perform convolution and spatial feature conversion processing on the first scale spatial feature map to extract a second scale spatial feature map, the second target condition vector having the same shape as the second scale spatial feature map;

use a third target condition vector to perform convolution and spatial feature conversion processing on the second scale spatial feature map to extract a third scale spatial feature map, the third target condition vector having the same shape as the third scale spatial feature map,

the first target condition vector, the second target condition vector, and the third target condition vector being obtained based on conversion processing of the target condition vector.

16 . The image processing device of claim 13 , wherein the processing module is further configured to:

convert the target condition vector into a ninth feature map having the same shape as an input feature map of a different scale space of the initial image;

fuse different ninth feature maps with the input feature map of the corresponding scale space into a tenth feature map;

perform convolution processing on different tenth feature maps to obtain corresponding feature tensors; and

use the feature tensors to perform spatial feature conversion processing on the input feature map corresponding to the scale space to obtain the third feature map set including a multi-scale space feature map.

17 . An electronic device comprising:

at least one processor;

and a memory communicatively coupled to the at least one processor, the memory storing an instruction set that, when executed, causes the processor to:

process an initial image to obtain a target condition vector, wherein the target condition vector represents degradation feature information of the initial image;

obtain filter weights in different scale spaces based on the target condition vector;

encode the initial image to obtain a multi-scale feature map;

process the multi-scale feature map using the different filter weights to obtain a target residual image; and

perform target processing on the initial image and the target residual image to obtain a target output image.

18 . The electronic device of claim 17 , wherein the instruction set further causes the processor to:

encode the initial image into a first feature map set, the first feature map set including feature maps of different scales;

convert the first feature map set into a second feature map set by using the corresponding filter weight, the second feature map set including feature maps of different scales; and

sequentially decode the feature maps in the second feature map set to obtain the target residual image.

19 . The electronic device of claim 18 , wherein the instruction set further causes the processor to:

perform convolution processing on a first feature map included in the first feature map set by using a first filter weight to obtain a fourth feature map in the second feature map set;

perform convolution processing on a second feature map included in the first feature map set by using a second filter weight to obtain a fifth feature map in the second feature map set; and

perform convolution processing on a third feature map included in the first feature map set by using a third filter weight to obtain a sixth feature map in the second feature map set;

wherein scale spaces corresponding to the first feature map, the second feature map, and the third feature map being different, and scale spaces corresponding to the fourth feature map, the fifth feature map, and the sixth feature map being different.

20 . The electronic device of claim 19 , wherein the instruction set further causes the processor to:

decode the sixth feature map to obtain a seventh feature map, the scale space corresponding to the seventh feature map being the same as the scale space of the fifth feature map;

accumulate the seventh feature map and the fifth feature map and performing decoding processing to obtain an eighth feature map, the scale space corresponding to the eighth feature map being the same as the scale space of the fourth feature map; and

perform residual processing after accumulating the eighth feature map and the fourth feature map to obtain the target residual image.