Image Compression Prediction via Color Variation Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image compression techniques face challenges in accurately predicting pixel values, leading to suboptimal compression rates due to significant differences between actual and predicted pixel values.
Innovation Solution
A method and system that predict pixel values based on reference pixel colors and selectively modify these predictions using a color-based correction unit to account for similarities in color variations within the image, thereby reducing differences between actual and predicted values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional spatial similarity based compression is used, then compression can be achieved, but prediction accuracy is insufficient leading to suboptimal compression rates
Solution Approach 1:
The patent changes the parameter used for prediction from simple spatial reference pixel values to color-difference based parameters. By calculating color differences (CbCr) between reference and current pixels, and using these color difference parameters for prediction, the system achieves better prediction accuracy that adapts to local color variations in the image, thereby improving compression rate.
2Measurement precision
If context-based correction is applied, then prediction accuracy improves, but device complexity increases
Solution Approach 1:
The patent extracts only the essential color difference information (CbCr components) from the full color data for prediction purposes. By taking out and using only the chrominance difference parameters rather than processing all color information, the system maintains improved prediction accuracy while avoiding the complexity of full context-based correction systems.
3Measurement precision
If color-based correction is implemented, then prediction errors are reduced, but computational complexity increases
Solution Approach 1:
The patent applies color-based correction locally by computing color differences only between specific reference pixels and current pixels in the processing order. By focusing computational effort on local color relationships rather than global image analysis, the system reduces prediction errors while keeping computational load manageable through localized processing.
Data Source
AI summary
A method, medium, and system compressing an image, and a method, medium, and system recovering an image. Values of colors of a pixel from among pixels making up an image are predicted from values of colors of a reference pixel corresponding to the pixel, and the predicted values of the colors of the pixel are corrected based on similarities in variations in color values in the image.