Combined Inter-Intra Prediction for Higher Image Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image encoding/decoding methods based on intra or inter prediction have lower compression ratios, which are inadequate for high-resolution and high-quality images, leading to increased transmission and storage costs.
Innovation Solution
An image encoding/decoding method that performs combined inter intra prediction, considering the structure and characteristics of blocks, by generating and combining prediction blocks using weights determined based on neighbor block prediction modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional intra prediction or inter prediction is used, then the encoding process is simple, but the compression ratio is low
Solution Approach 1:
The patent combines inter prediction and intra prediction into a unified combined prediction process. The prediction unit generates both inter prediction samples (from reference pictures) and intra prediction samples (from current picture neighbors), then combines them using weighting factors to produce the final prediction. This merging of prediction modes enables the system to achieve higher compression ratios by leveraging the advantages of both inter and intra prediction for each block.
Solution Approach 2:
The patent introduces dynamic weighting factors that can be adjusted based on block characteristics and prediction quality. The weighting factors are not fixed but can be adapted to optimize the combination of inter and intra prediction samples. This dynamic adjustment allows the system to balance the complexity of processing with the benefits of combined prediction, improving compression efficiency while managing computational requirements.
2Productivity
If combined inter intra prediction is implemented, then encoding efficiency is improved, but processing complexity increases
Solution Approach 1:
The patent divides the prediction process into distinct segments: generating inter prediction samples, generating intra prediction samples, and combining them. Each segment can be processed independently using dedicated algorithms and data structures. The prediction unit is structured to handle inter and intra prediction separately before merging, which reduces the immediate complexity of any single operation while achieving the benefits of combined prediction.
Solution Approach 2:
The patent applies different prediction methods and weighting factors to different blocks based on their specific characteristics. Blocks with high motion activity might rely more on inter prediction, while blocks with smooth regions might benefit more from intra prediction. This local adaptation allows the system to optimize encoding efficiency for each block type while managing overall processing complexity through targeted computation.
3Manufacturing precision
If high resolution and quality images are transmitted, then image quality is improved, but transmission cost and storage cost increase
Solution Approach 1:
The patent changes the prediction parameters and weighting factors adaptively based on image content and block characteristics. By optimizing the combination of inter and intra prediction samples, the system achieves better compression ratios that preserve image quality while reducing the amount of data that needs to be transmitted and stored. This parameter adaptation enables high-quality image transmission with lower data requirements.
Data Source
AI summary
Disclosed herein is a method of decoding an image. The method includes generating an inter prediction block by performing inter prediction with respect to a current block, generating an intra prediction block by performing intra prediction with respect to the current block, determining a first weight and a second weight, and generating a final prediction block by respectively applying the first weight and the second weight to the inter prediction block and the intra prediction block. The generating of the intra prediction block includes generating the intra prediction block using a predefined intra prediction mode.