Frame Buffer Compression via Local Multiscale Transform
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video processing devices face challenges in reducing the size of internal memory and bandwidth requirements due to high data transfer rates and the need for large external DRAM, with existing compression techniques offering limited compression ratios and image quality trade-offs, especially when requiring constant bit rate control.
Innovation Solution
A method involving a local multiscale transform to generate coefficient blocks, distributing coefficients into groups with a common exponent and mantissas for quantization, allowing for scalable and efficient compression and decompression of video signals, enabling selective access to frame data with minimal overhead and reduced memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If compression techniques are applied to reduce memory size and bandwidth requirements, then memory requirements are reduced, but image quality may deteriorate and compression ratio control becomes difficult
Solution Approach 1:
The patent applies a local multiscale transform that converts image data into coefficient blocks with different scale levels. By adjusting the number of scales and the quantization precision of mantissas, the system can dynamically control compression ratio while maintaining image quality. The exponent-mantissa representation allows flexible parameter adjustment to achieve constant bit rate compression.
Solution Approach 2:
The patent divides the image into coefficient blocks and further segments them into groups with common exponents and individual mantissas. This segmentation enables selective compression where different regions can have different compression levels, allowing the system to maintain high image quality in important regions while achieving higher compression in less critical areas, thus resolving the contradiction between compression ratio and image quality.
2Productivity
If high compression ratios are achieved, then memory bandwidth requirements are reduced, but image quality and constant bit rate control become problematic
Solution Approach 1:
The exponent-mantissa representation system allows dynamic adjustment of compression parameters. By changing the number of bits allocated to mantissas and the scaling factors (exponents), the system can precisely control the bit rate while maintaining high compression ratios. This parameter flexibility enables constant bit rate control even at high compression levels.
Solution Approach 2:
The patent implements a dynamic compression system where the transform depth, quantization levels, and exponent-mantissa allocation can be adjusted in real-time based on content complexity and desired bit rate. This dynamic adaptation allows the system to maintain constant bit rate control while achieving high compression ratios, as the compression parameters are continuously optimized rather than fixed.
3Volume of stationary object
If sequential processing with tile splitting is used to reduce internal buffer size, then line buffer size is reduced, but data rate increases due to overlapping requirements
Solution Approach 1:
The local multiscale transform processes image data in localized coefficient blocks rather than requiring global tile processing. Each block can be independently compressed and stored with its own exponent-mantissa representation, eliminating the need for overlapping tiles and the associated increased data rates. This local processing approach reduces internal buffer requirements without increasing bandwidth requirements.
Data Source
AI summary
For compressing a video signal, a local multiscale transform is applied to a frame of the video signal to obtain coefficient blocks. The coefficients of each block are distributed into a plurality of coefficient groups, and for at least one of the groups, a common exponent is determined for encoding the coefficients of the group, and respective mantissas are determined for quantizing the coefficients of the group in combination with the common exponent. Coding data including each exponent determined for a coefficient group and the mantissas quantizing the coefficients of the group in combination with this exponent are stored in an external frame buffer.