Electronic apparatus and control method thereof

1. An electronic apparatus comprising:

a storage configured to store a compression rate network model configured to identify a compression rate applied to an image block from among a plurality of compression rates, and a plurality of compression noise removing network models configured to remove compression noise for each of the plurality of compression rates; and

a processor configured to:

receive a decoded moving picture including a sequence of a plurality of decoded frames,

divide a frame of the plurality of decoded frames into a plurality of image blocks,

obtain a predicted compression rate of each of the plurality of image blocks included in the frame of the decoded moving picture based on the compression rate network model,

determine most frequent compression rate among the plurality of obtained predicted compression rates for the plurality of image blocks of the frame as a predicted compression rate of the frame, and

remove compression noise of the frame based on a compression noise removing network model corresponding to the compression rate of the frame from among the plurality of compression noise removing network models,

wherein the compression rate network model is obtained by learning image characteristics of a plurality of restored image blocks corresponding to the plurality of compression rates through a first artificial intelligence algorithm,

wherein the plurality of restored image blocks are generated by encoding a plurality of original image blocks, and decoding the encoded plurality of original image blocks, and

wherein the plurality of compression noise removing network models are obtained by learning a relation between the plurality of original image blocks and the plurality of restored image blocks through a second artificial intelligence algorithm.

2. The electronic apparatus of claim 1 , wherein the processor is configured to divide each of the plurality of frames in a preset number of frames before and after the frame into a plurality of additional image blocks, to obtain an additional predicted compression rate of each of the plurality of additional image blocks based on the compression rate network model, and to obtain a compression rate of the frame based on the plurality of compression rates and the plurality of additional compression rates.

3. The electronic apparatus of claim 1 , wherein the processor is configured to calculate dispersion for a pixel value of each of the plurality of image blocks in the frame, to obtain image blocks in a preset number among the plurality of image blocks based on the calculated dispersion, to obtain a compression rate of each of the obtained image blocks in the preset number based on the compression network model, and to obtain a compression rate of the frame based on the plurality of obtained compression rates.

4. The electronic apparatus of claim 1 , wherein the processor is configured to remove compression noise of each of the plurality of image blocks based on the compression noise removing network model, and to obtain a frame in which compression noise has been removed by combining the plurality of image blocks in which compression noise has been removed.

5. The electronic apparatus of claim 4 , wherein the processor is configured to obtain a second image block and a third image block corresponding to a first image block, the first image block being one of the plurality of image blocks in frames before and after the frame, and to remove compression noise of the first image block using the second image block and the third image block.

6. The electronic apparatus of claim 5 , wherein the processor is configured to perform at least one convolution operation for each of the first image block, the second image block, and the third image block based on the compression noise removing network model, and to combine the first image block, the second image block, and the third image block for which the convolution operation has been performed, and to remove compression noise of the first image block by performing at least one convolution operation for the combined image blocks.

7. The electronic apparatus of claim 5 , wherein the processor is configured to calculate a first peak signal-to-noise ratio (PSNR) of the first image block and the second image block, and a second PSNR of the first image block and the third image block, and based on at least one of the first PSNR and the second PSNR being smaller than a preset value, input the first image block in the compression noise removing network model in place of an image block corresponding to a PSNR smaller than the preset value between the second image block and the third image block.

8. The electronic apparatus of claim 4 , wherein the processor is configured to combine the plurality of image blocks in which compression noise has been removed by overlapping the blocks based on a weight based on a distance from a boundary.

9. The electronic apparatus of claim 1 , wherein the processor is configured to divide the frame into a chrominance component and a luminance component, to remove compression noise of the luminance component based on the compression noise removing network model, and to remove compression noise of the frame by combining the chrominance component and the luminance component in which compression noise has been removed, and

wherein the plurality of compression noise removing network models are obtained by learning a relation between a luminance component of the plurality of original image blocks and a luminance component of the plurality of restored image blocks through a second artificial intelligence algorithm.

10. The electronic apparatus of claim 1 , wherein the compression network model and the plurality of compression noise removing network models have a deep learning network structure.

11. A method controlling of an electronic apparatus, comprising:

receiving a decoded moving picture including a sequence of a plurality of decoded frames;

dividing a frame of the plurality of decoded frames into a plurality of image blocks;

obtaining a predicted compression rate of each of the plurality of image blocks included in the frame of the decoded moving picture based on a compression rate network model configured to identify a compression rate applied to an image block from among a plurality of compression rates;

determining most frequent compression rate among the plurality of obtained predicted compression rates for the plurality of image blocks of the frame as a predicted compression rate of the frame, and

removing compression noise of the frame based on a compression noise removing network model corresponding to the compression rate of the frame from among a plurality of compression noise removing network models configured to remove compression noise for each of the plurality of compression rates,

wherein the compression rate network model is obtained by learning image characteristics of a plurality of restored image blocks corresponding to the plurality of compression rates through a first artificial intelligence algorithm,

wherein the plurality of restored image blocks are generated by encoding a plurality of original image blocks, and decoding the encoded plurality of original image blocks, and

wherein the plurality of compression noise removing network models are obtained by learning a relation between the plurality of original image blocks and the plurality of restored image blocks through a second artificial intelligence algorithm.

12. The method of controlling an electronic apparatus of claim 11 ,

wherein determining comprises:

dividing each of the plurality of frames in a preset number of frames before and after the frame into a plurality of additional image blocks;

obtaining an additional predicted compression rate of each of the plurality of additional image blocks based on the compression rate network model; and

obtaining a compression rate of the frame based on the plurality of compression rates and the plurality of additional compression rates.

13. The method of controlling an electronic apparatus of claim 11 ,

wherein obtaining a compression rate of each of the plurality of image blocks comprises:

calculating dispersion for a pixel value of each of the plurality of image blocks in the frame;

obtaining image blocks in a preset number among the plurality of image blocks based on the calculated dispersion; and

obtaining a compression rate of each of the obtained image blocks in the preset number based on the compression network model, and

wherein determining comprises:

obtaining a compression rate of the frame based on the compression rate of each of the obtained image blocks in a preset number.

14. The method of controlling an electronic apparatus of claim 11 ,

wherein removing compression noise comprises:

removing compression noise of each of the plurality of image blocks based on the compression noise removing network model; and

obtaining a frame where compression noise has been removed by combining the plurality of image blocks where compression noise has been removed.

15. The method of controlling an electronic apparatus of claim 14 ,

wherein removing compression noise further comprises:

obtaining a second image block and a third image block corresponding to a first image block, the first image block being one of the plurality of image blocks in frames before and after the frame; and

removing compression noise of the first image block using the second image block and the third image block.

16. The method of controlling an electronic apparatus of claim 15 ,

wherein removing compression noise of the first image block comprises:

performing at least one convolution operation for each of the first image block, the second image block, and the third image block based on the compression noise removing network model;

combining the first image block, the second image block, and the third image block for which the convolution operation has been performed; and

removing compression noise of the first image block by performing at least one convolution operation for the combined image blocks.

17. The method of controlling an electronic apparatus of claim 15 ,

wherein removing compression noise further comprises:

calculating a first peak signal-to-noise ratio (PSNR) of the first image block and the second image block, and a second PSNR of the first image block and the third image block; and

based on at least one of the first PSNR and the second PSNR being smaller than a preset value, inputting the first image block in the compression noise removing network model in place of an image block corresponding to a PSNR smaller than the preset value between the second image block and the third image block.

18. The method controlling of an electronic apparatus of claim 14 , further comprising: combining the plurality of image blocks in which compression noise has been removed by overlapping the image blocks based on a weight based on a distance from a boundary.

19. The method of controlling an electronic apparatus of claim 11 ,

wherein removing compression noise comprises:

dividing the frame into a chrominance component and a luminance component;

removing compression noise of the luminance component based on the compression noise removing network model; and

removing compression noise of the frame by combining the chrominance component and the luminance component in which compression noise has been removed, and

wherein the plurality of compression noise removing network models are obtained by learning a relation between a luminance component of the plurality of original image blocks and a luminance component of the plurality of restored image blocks through a second artificial intelligence algorithm.

20. The method of controlling an electronic apparatus of claim 11 , wherein the compression network model and the plurality of compression noise removing network models have a deep learning network structure.