Image Compression Block Prediction Using K-Means Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image compression techniques, such as those used in MPEG and JPEG standards, often sacrifice fine details to achieve higher compression rates, leading to lossy compression that may not be optimal for all types of digital content, and there is a need for improved methods to efficiently encode and decode images while preserving quality.
Innovation Solution
A computer-implemented method and system that identifies a target block in an image, selects candidate source blocks for prediction, ranks them based on residual information and position, and uses the best predictor block for encoding, storing the compressed image with minimal information loss, and decodes images using the same reference blocks to reconstruct the original image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If lossy compression techniques are used to achieve higher compression rates, then the compression ratio is improved, but the image quality and fine details are lost
Solution Approach 1:
The image is divided into multiple blocks, and each block is processed independently using block prediction. This segmentation allows the encoder to select the most appropriate reference block for each target block, preserving local details while achieving compression. The patent divides the image into 16x16 pixel blocks and uses intra-block prediction by selecting reference blocks from neighboring areas.
Solution Approach 2:
The patent changes the prediction parameter by selecting reference blocks based on their position and residual information rather than using fixed prediction modes. The encoder evaluates multiple candidate reference blocks and selects the one that minimizes the residual, adapting the prediction parameter dynamically to preserve image quality while achieving compression.
2Productivity
If block prediction with candidate source blocks is used, then the compression efficiency is improved, but the computational complexity increases
Solution Approach 1:
Instead of evaluating all possible reference blocks in the entire image, the patent limits the search to a set of candidate source blocks that are spatially close to the target block. This partial action approach evaluates only the most relevant candidates, improving compression efficiency while keeping computational complexity manageable.
Solution Approach 2:
The patent applies local quality by selecting reference blocks from local neighboring areas rather than searching the entire image. Each block uses local prediction from adjacent blocks, which reduces the search space and computational complexity while maintaining prediction accuracy for local image regions.
3Measurement precision
If more candidate source blocks are evaluated for prediction, then the prediction accuracy is improved, but the encoding time increases
Solution Approach 1:
The patent evaluates only a limited set of candidate source blocks that are most likely to provide good prediction, rather than exhaustively searching all possible blocks. This partial evaluation approach achieves sufficient prediction accuracy while significantly reducing encoding time compared to exhaustive search methods.
Solution Approach 2:
The patent performs preliminary ranking of candidate source blocks based on their position and residual information before selecting the final reference block. This preliminary action allows the encoder to quickly identify the best candidate without exhaustive evaluation, improving prediction accuracy while controlling encoding time.
Data Source
AI summary
Compression of an image is performed based on prediction of target blocks of an image from candidate source blocks of the image. Heuristics are used for identifying the candidate source blocks, for example, source blocks are selected from within a cluster of similar blocks obtained by K-means clustering. For each target block, a region adjacent to the target block is identified and a set of candidate source blocks along with candidate source regions adjacent to the candidate source blocks are identified. The candidate source regions are ranked based on the differences between the candidate source regions and the target source region. Each candidate source block is described using its rank and residual information describing differences between the candidate source block and the target block. The candidate source block that can be described using a minimum amount of information is selected for predicting the target block.