Rectangular Block Intra Prediction for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding methods, such as H.264, face inefficiencies in prediction accuracy due to using relatively distant pixels for intra prediction in square block units, leading to degraded compression efficiency.
Innovation Solution
The method involves predicting a current block's pixels from adjacent blocks using a rectangular shape aligned with the prediction direction, generating a rectangular prediction block, and then transforming and encoding the residual block into a square format for efficient compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If square block units are used for intra prediction in conventional video coding, then the encoding process is simplified and standardized, but prediction accuracy deteriorates because relatively distant pixels are used for prediction
Solution Approach 1:
The current block is divided into multiple sub-blocks along the prediction direction, with each sub-block using its own closest adjacent pixels for prediction. This segmentation allows each sub-block to perform prediction independently using locally proximal pixels, improving overall prediction accuracy without requiring complex global optimization
Solution Approach 2:
Different regions (sub-blocks) of the current block use different prediction reference pixels based on their local position and direction. Each sub-block selects reference pixels that are locally closest to it, ensuring that each local region uses the most appropriate reference pixels for its specific spatial context, thereby improving local prediction accuracy
2Productivity
If square block units are used for intra prediction, then processing is more regular and efficient, but compression efficiency deteriorates due to lower prediction accuracy
Solution Approach 1:
By segmenting the block into sub-blocks that each use their own closest reference pixels, the prediction accuracy for each sub-block is improved, leading to smaller residuals and better compression efficiency in the transformed domain
Solution Approach 2:
The patent introduces a new dimension of consideration by evaluating prediction accuracy based on pixel proximity in the spatial domain rather than just block-based processing. This dimensional shift from block-centric to pixel-proximity-centric prediction enables better prediction while maintaining processing efficiency through systematic sub-block processing
Data Source
AI summary
Video encoding/decoding method and apparatus using a block transformation are disclosed. The apparatus encodes the video through predicting each pixel in a current block of the video by using one or more adjacent pixels, which are encoded prior to encoding the current block and have the closest proximity to the pixels respectively in a predetermined prediction direction, and wherein the current block is sized L×M with L being unequal to M. It has become possible to predict the pixels in the block starting from closest one of the pixels in the prediction direction in order to provide more accurate block predictions and then raise the compression efficiency that leaps to more efficient video encoding/decoding performance.