Block-Size Prediction Strategy for Image Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In image processing, the high memory bandwidth required for decoding images in advanced coding methods like AVC and HEVC leads to increased costs due to the need for efficient prediction image generation and storage, particularly for large picture frames like Ultra High Definition.
Innovation Solution
An image processing apparatus and method that sets identification information for block sizes and prediction methods, generating prediction images based on this information, and transmitting coded streams with this information to optimize memory usage and reduce costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If bi-prediction is used for all block sizes to improve coding efficiency, then image quality and compression ratio are improved, but memory bandwidth requirements and apparatus costs increase
Solution Approach 1:
The patent applies different prediction methods to different block sizes. Specifically, bi-prediction is applied to large blocks (16x16 or larger) while uni-prediction is applied to small blocks (8x8 or smaller). This local differentiation optimizes coding efficiency for large blocks while reducing memory bandwidth consumption for small blocks, thereby resolving the contradiction between coding efficiency and memory bandwidth requirements.
2Productivity
If bi-prediction is used for small blocks to improve compression ratio, then coding efficiency is improved, but the complexity of prediction image generation increases
Solution Approach 1:
The patent restricts bi-prediction to large blocks (16x16 or larger) and uses uni-prediction for small blocks (8x8 or smaller). This approach maintains high compression ratio for large blocks where bi-prediction provides significant benefits, while avoiding the complexity of bi-prediction in small blocks where the overhead outweighs the benefits.
3Productivity
If advanced prediction methods are used for all blocks to improve coding efficiency, then image quality is improved, but apparatus costs increase due to higher memory bandwidth requirements
Solution Approach 1:
The patent implements a block-size-dependent prediction strategy where bi-prediction is used for large blocks (16x16 or larger) to maintain high image quality, while uni-prediction is used for small blocks (8x8 or smaller) to reduce memory bandwidth requirements and apparatus costs. This selective approach ensures high image quality where it matters most while controlling manufacturing costs.
Data Source
AI summary
The present technology relates to an image processing apparatus and method capable of preventing an increase in a cost of the apparatus. A setting unit sets identification information for identifying a correspondence relationship between a size of a block of an image and changing of a prediction method applied to the block, and an inter-prediction unit generates a prediction image according to the identification information. An encoder codes the block by using the prediction image so as to generate a coded stream. In addition, the encoder transmits the coded stream and the identification information. The present technology is applicable to, for example, a case of coding/decoding an image, and the like.