Image Block Prediction Optimization for JPEG Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current JPEG image compression techniques have limited compression efficiency due to restrictive prediction methods that only consider DC components, leading to suboptimal compression results.
Innovation Solution
The method optimizes prediction by computing prediction errors for various methods (DC, horizontal, vertical, vertical left, and vertical right) across blocks, selecting the optimal method based on minimum summation of prediction errors and encoding these through DCT, Zigzag scan, and Huffman Coding to minimize output bits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If DC prediction method is used for image compression, then implementation is simple, but compression efficiency is limited
Solution Approach 1:
The patent applies dynamics by transitioning from a static DC prediction method to a dynamic selection mechanism that chooses among multiple prediction methods (DC, horizontal, vertical, vertical left, vertical right) based on the actual image content characteristics. The system dynamically adapts the prediction approach for each block to maximize compression efficiency while maintaining implementation feasibility through standardized algorithmic choices.
Solution Approach 2:
The patent changes the parameter of prediction method selection by introducing multiple prediction modes with different directional assumptions. Instead of using a single DC prediction parameter, the system varies the prediction approach (horizontal, vertical, diagonal directions) to match the local image structure, thereby improving compression efficiency without significantly complicating the overall implementation.
2Productivity
If multiple prediction methods are computed and evaluated, then compression efficiency improves, but computational complexity increases
Solution Approach 1:
The patent applies partial action by computing prediction errors for multiple prediction methods but only fully processing the optimal one for final encoding. The system performs excessive computation in the evaluation phase (computing all prediction errors) but limits the final encoding to the single best method, thereby balancing compression efficiency improvement with acceptable computational complexity.
Solution Approach 2:
The patent uses preliminary action by pre-computing prediction errors for all candidate methods before final encoding. This preliminary evaluation phase identifies the optimal prediction method in advance, allowing the main encoding process to proceed efficiently with only the selected method, thus managing computational complexity while achieving improved compression.
3Manufacturing precision
If prediction error minimization is prioritized, then compression ratio improves, but number of bits for encoding increases
Solution Approach 1:
The patent applies feedback by using the computed prediction errors to guide the selection of prediction methods and subsequent encoding parameters. The system feeds back the error magnitude information to adjust the encoding strategy, choosing methods that achieve good prediction accuracy while considering the bit cost, thereby balancing prediction precision with compression ratio.
Solution Approach 2:
The patent changes parameters by adjusting the prediction method selection and encoding parameters based on the observed prediction errors. When prediction errors are small, the system can use more aggressive compression; when errors are large, it adjusts to maintain quality. This dynamic parameter adjustment balances prediction accuracy requirements with bit rate constraints.
Data Source
AI summary
A technique for optimizing a prediction method of samples in blocks of an image is provided. The image includes a first block, a second block, a third block, and a fourth block, each of the blocks include 8×8 blocks and form one Macro block. The method includes performing a prediction of the second block, the third block and the fourth block by performing at least one of prediction methods. A prediction error per block (Pe) is computed for each prediction method that is performed to predict the second block, the third block, and the fourth block. The prediction error per block (Pe) equals an original block value minus a predicted block value. An optimal prediction method is chosen from each of the prediction methods performed that results in minimum Σ|Pe| pixels per block (summation on pixels per block). The optimal prediction method and the Pe for each block are stored.