Reference Image Memory Compression for Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image reproduction techniques for decoding moving picture streams suffer from temporal accumulation of compression distortion, leading to increased memory bandwidth requirements and power consumption, particularly in small household cameras, due to the need for high-performance memory and irreversible transforms that introduce noise in the decoded images.
Innovation Solution
An image reproducing method that estimates traffic to the reference image memory on a picture-by-picture basis and determines the compression mode for storing reference images based on this traffic, reducing unnecessary compression and thus minimizing temporal accumulation of compression distortion, without the need for high-performance memory.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If high-performance memory (plurality of DRAMs or high-speed DRAM) is used to secure high-frequency memory bandwidth for reference image storage, then decoding performance is improved, but packaging costs and power consumption increase
Solution Approach 1:
The patent performs preliminary action by conducting a first decoding process before the actual second decoding to estimate the traffic to the reference image memory in advance. This estimation allows the system to determine the appropriate compression mode beforehand, avoiding the need for high-performance memory while ensuring the memory bandwidth requirements are met for the actual decoding operation.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting the compression mode (compression ratio) of reference images based on the estimated traffic. By changing the compression parameter adaptively - using higher compression when traffic is low and lower compression when traffic is high - the system optimizes memory bandwidth usage without requiring high-performance memory, thereby reducing power consumption.
2Reliability
If high-performance memory is used to store reference images, then image reproduction quality is improved, but manufacturing costs increase
Solution Approach 1:
The first decoding process serves as a preliminary action that estimates memory traffic before the actual decoding. This advance estimation enables the system to configure appropriate compression levels for reference images, ensuring reproduction quality is maintained while avoiding the need for expensive high-performance memory hardware.
Solution Approach 2:
The compression mode parameter is changed adaptively based on traffic estimation. By adjusting the compression ratio according to actual needs rather than using fixed high-compression settings, the system maintains image reproduction quality while using standard, cost-effective memory components.
3Quantity of substance
If irreversible transform is applied to compress reference images, then memory bandwidth is reduced, but compression distortion accumulates over time
Solution Approach 1:
The first decoding process performs preliminary traffic estimation before the actual second decoding. This advance knowledge allows the system to select the appropriate compression mode, applying irreversible compression only when traffic is sufficiently low, thereby preventing excessive compression distortion accumulation while still achieving memory bandwidth reduction.
Solution Approach 2:
The patent introduces dynamics by making the compression mode adaptable rather than static. The compression ratio changes dynamically based on the estimated traffic for each picture, allowing the system to balance between memory bandwidth reduction and distortion accumulation in real-time according to actual decoding requirements.
4Quantity of substance
If compression mode is increased for all reference images, then memory bandwidth is reduced, but image quality deteriorates due to distortion accumulation
Solution Approach 1:
The first decoding process provides preliminary traffic estimation that guides the compression mode selection. This advance planning ensures that high compression is applied only when appropriate, preventing quality deterioration while achieving memory bandwidth reduction where possible.
Solution Approach 2:
The compression mode parameter is changed adaptively based on traffic estimation rather than applying a fixed high compression ratio to all images. This selective parameter adjustment maintains image quality by using lower compression when traffic requires it, while still achieving bandwidth reduction when conditions allow.
Data Source
AI summary
A image reproducing device decodes a moving picture stream having been subjected to inter-picture prediction encoding with a reference image memory accessed. A decoding processor performs variable length decoding of the moving picture stream on a picture-by-picture basis, and outputs an encoding parameter. A traffic analyzer estimates traffic to the reference image memory on the picture-by-picture basis based on the output encoding parameters. A reference image compression controller determines a compression mode in storing a reference image in the reference image memory based on the estimated traffic.