Image processing method, apparatus, and system
View Patent ↗An image processing method, apparatus, and system are provided. The image processing method includes: obtaining an initial image; converting the initial image from one type to another based on an image feature of the initial image, to obtain a target image; and presenting the target image, or analyzing and processing the target image. The initial image can be processed based on the image feature of the initial image, to obtain another type of image.
1. An image processing method comprising:
obtaining an initial image, wherein the initial image is of a first type;
obtaining a target image based on a global image feature and a local image feature of the initial image, wherein the target image is of a second type; and
generating for display an interface presenting the target image or analyzing and processing the target image,
wherein the global image feature of the initial image is determined based on a second feature map of the initial image that is obtained through performing adaptive pooling on a first feature map of the initial image,
wherein the first feature map of the initial image is obtained through convolution processing on the initial image, and
wherein the second feature map of the initial image is a fixed-size feature map.
2. The image processing method according to claim 1 , wherein the local image feature of the initial image is obtained through convolution processing on the initial image.
3. The image processing method according to claim 1 , wherein the obtaining the global image feature of the initial image comprises:
obtaining the first feature map of the initial image, wherein the first feature map is obtained after L th convolution processing is performed on the initial image, a size of the first feature map is M×N, and L, M, and N are positive integers;
performing adaptive pooling on the first feature map based on the size of the first feature map to obtain the second feature map, wherein a size of the second feature map is R×S, and R and S are preset fixed values and positive integers; and
determining the global image feature of the initial image based on the second feature map.
4. The image processing method according to claim 3 , wherein the performing adaptive pooling on the first feature map based on the size of the first feature map to obtain the second feature map comprises:
performing pooling on the first feature map by using a sliding window, to obtain the second feature map, wherein a size of the sliding window and a sliding step of the sliding window are determined based on the sizes of the first feature map and the second feature map.
5. The image processing method according to claim 4 , wherein the size of the sliding window and the sliding step of the sliding window are determined according to the following formulas:
s _ h=M//R;
s _ w=N//S;
k _ h=M −( R− 1)* s _ h ; and
k _ w=N −( S− 1)* s _ w , wherein
the size of the sliding window is (k_h, k_w), the sliding step of the sliding window is (s_h, s_w), and//represents exact division.
6. The image processing method according to claim 1 , wherein the obtaining the target image based on a global image feature and a local image feature of the initial image comprises:
obtaining a final feature map of the initial image based on the global image feature and the local image feature of the initial image; and
performing upsampling multiple times on the final feature map of the initial image, to obtain the target image.
7. The image processing method according to claim 6 , wherein the obtaining a final feature map of the initial image based on the global image feature and the local image feature of the initial image comprises:
splicing the global image feature and the local image feature of the initial image, to obtain the final feature map of the initial image.
8. The image processing method according to claim 1 , wherein the initial image is a standard dynamic range (SDR) image, and the target image is a high dynamic range (HDR) image.
9. The image processing method according to claim 1 , wherein the method comprises analyzing and processing the target image, and the analyzing and processing the target image comprises:
performing image classification or recognition processing on the target image, to obtain a processing result of the target image.
10. A controller, comprising:
at least one processor; and
a non-transitory computer-readable storage medium coupled to the at least one processor and storing programming instructions that, when executed by the at least one processor, instruct the controller to perform an image processing method including the operations comprising:
obtaining an initial image, wherein the initial image is of a first type;
obtaining a target image based on a global image feature and a local image feature of the initial image, wherein the target image is of a second type; and
generating for display an interface presenting the target image or analyzing and processing the target image,
wherein the global image feature of the initial image is determined based on a second feature map of the initial image that is obtained through performing adaptive pooling on a first feature map of the initial image,
wherein the first feature map of the initial image is obtained through convolution processing on the initial image, and
wherein the second feature map of the initial image is a fixed-size feature map.
11. The controller according to claim 10 , wherein the local image feature of the initial image is obtained through convolution processing on the initial image.
12. The controller according to claim 10 , wherein the programming instructions, when executed by the at least one processor, further instruct the controller to perform operations comprising:
obtaining the first feature map of the initial image, wherein the first feature map is obtained after L th convolution processing is performed on the initial image, a size of the first feature map is M×N, and L, M, and N are positive integers;
performing adaptive pooling on the first feature map based on the size of the first feature map to obtain the second feature map, wherein a size of the second feature map is R×S, and R and S are preset fixed values and positive integers; and
determining the global image feature of the initial image based on the second feature map.
13. The controller according to claim 12 , wherein the programming instructions, when executed by the at least one processor, further instruct the controller to perform operations comprising:
performing pooling on the first feature map by using a sliding window, to obtain the second feature map, wherein a size of the sliding window and a sliding step of the sliding window are determined based on the sizes of the first feature map and the second feature map.
14. The controller according to claim 13 , wherein the size of the sliding window and the sliding step of the sliding window are determined according to the following formulas:
s _ h=M//R;
s _ w=N//S;
k _ h=M −( R− 1)* s _ h ; and
k _ w=N −( S− 1)* s _ w , wherein
the size of the sliding window is (k_h, k_w), the sliding step of the sliding window is (s_h, s_w), and//represents exact division.
15. The controller according to claim 10 , wherein the programming instructions, when executed by the at least one processor, further instruct the controller to perform operations comprising:
obtaining a final feature map of the initial image based on the global image feature and the local image feature of the initial image; and
performing upsampling multiple times on the final feature map of the initial image, to obtain the target image.
16. The controller according to claim 15 , wherein the programming instructions, when executed by the at least one processor, further instruct the controller to perform operations comprising:
splicing the global image feature and the local image feature of the initial image, to obtain the final feature map of the initial image.
17. The controller according to claim 10 , wherein the second feature map is obtained through adaptive pooling performed on the first feature map using a sliding window and using average pooling.
18. The controller according to claim 10 , wherein the second feature map is obtained through adaptive pooling performed on the first feature map using a sliding window and using maximum pooling.
19. The method according to claim 1 , wherein the second feature map is obtained through adaptive pooling performed on the first feature map using a sliding window and using average pooling.
20. The method according to claim 1 , wherein the second feature map is obtained through adaptive pooling performed on the first feature map using a sliding window and using maximum pooling.