ASTC Encoder PSNR Computation via Decimated Weight Block Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing texture compression techniques require a decoder to compute peak signal to noise ratio (PSNR), which consumes additional silicon hardware area and power.
Innovation Solution
The method involves identifying pixel positions in a coded block corresponding to an index in a decimated weight block and extracting error values using a microprocessor, eliminating the need for a decoder path by employing a weight-on-weight method for PSNR computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a decoder is used to compute PSNR by decoding the encoded image, then PSNR computation can be performed, but additional silicon hardware area and power consumption increase
Solution Approach 1:
The patent extracts only the essential error computation functionality from the full decoder system. Instead of implementing a complete decoder that reconstructs entire images, the invention extracts and implements only the weighted error calculation mechanism using the decimated weight block and pixel position mapping, thereby obtaining PSNR computation capability with minimal hardware resources
Solution Approach 2:
The patent creates a simplified copy of the error computation process that mirrors the essential functionality of the full decoder without requiring the complete decoding pipeline. By copying only the weighted error calculation logic and using the pre-computed decimated weight block, the system achieves PSNR computation without replicating the entire decoder hardware structure
2Measurement precision
If a decoder is used to compute PSNR, then PSNR can be calculated, but power consumption increases significantly
Solution Approach 1:
The patent extracts only the essential error computation functionality from the full decoder system. Instead of implementing a complete decoder that reconstructs entire images, the invention extracts and implements only the weighted error calculation mechanism using the decimated weight block and pixel position mapping, thereby obtaining PSNR computation capability with minimal hardware resources
Solution Approach 2:
The patent applies partial action by computing PSNR for only the essential pixel positions that map to the decimated weight block, rather than computing errors for all pixels in the original image. This selective error computation at critical positions achieves sufficient PSNR measurement without the full power cost of decoding the entire image
3Measurement precision
If block-based encoding is used with full decoding, then accurate error difference can be obtained, but device complexity increases
Solution Approach 1:
The patent segments the error computation process by dividing the weight block into a decimated weight block with reduced resolution. This segmentation allows the system to compute errors at representative pixel positions that map to the decimated weights, achieving block-based error analysis without requiring full-resolution decoding hardware for every block
Solution Approach 2:
The patent extracts only the essential error computation functionality from the full decoder system. Instead of implementing a complete decoder that reconstructs entire images, the invention extracts and implements only the weighted error calculation mechanism using the decimated weight block and pixel position mapping, thereby obtaining PSNR computation capability with minimal hardware resources
Data Source
AI summary
A method of extracting error for peak signal to noise ratio (PSNR) computation in an adaptive scalable texture compression (ASTC) encoder. The method includes identifying a pixel positions of a pixel in a coded block. The pixel position corresponds to an index in a decimated weight block. The method extracts an error value corresponding to the identified pixel position for said PSNR computation in said ASTC encoder.