An inter-frame prediction method based on deep neural network
A deep neural network and inter-frame prediction technology, applied in the field of inter-frame prediction, can solve problems such as insufficient accuracy, achieve the effects of improving accuracy, improving coding efficiency, and reducing time-domain redundancy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0041] A kind of inter-frame prediction method based on deep neural network, the process of described method is, as figure 1 Shown:
[0042] Step 1: Obtain the surrounding adjacent pixels of the current block, the reference block and the surrounding adjacent pixels of the reference block, the current block and the reference block are rectangular areas or non-rectangular areas; when the current block and the reference block are rectangular areas , the size of the current block and the reference block is W*H, W is the width of the current block and the reference block, and H is the height of the current block and the reference block;
[0043] Step 2: Input the surrounding adjacent pixels of the current block, the reference block and the surrounding adjacent pixels of the reference block obtained in step 1 into the deep neural network, and learn to obtain the relationship between the current block and the reference block, or learn to obtain the reference block and the relationsh...
Embodiment 2
[0069] The difference between embodiment 2 and embodiment 1 is that the process of obtaining a more accurate prediction block in the second step is:
[0070] Step 1: Input the reference block obtained in step 1 and the surrounding adjacent pixels of the reference block into a neural network to learn the relationship between the reference block and the surrounding adjacent pixels of the reference block. The neural network consists of fully connected layers, convolutional layers, or a combination of both;
[0071] The second step: input the surrounding adjacent pixels of the current block obtained in step 1 and the relationship obtained in the first step into a neural network, and learn to obtain a more accurate prediction block of the current block. The neural network consists of a fully connected layer, Convolutional layers or a combination of both.
Embodiment 3
[0073] The difference between embodiment 3 and embodiment 1 is that, in the inter-frame prediction method in the hybrid video codec system of this embodiment, the deep neural network of the first step, the second step and the third step in step 2 can be passed through simple The transformations are integrated into a deep neural network. In principle, the distinction between the first step, the second step, and the third step is for the convenience of description, and they are distinguished according to their functions. During training and deployment, the entire network is in an end-to-end manner, so conceptually distinguishing network modules is a special case of Embodiment 1.
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


