Quantization Control for Parallel Decoder Throughput

Solution Overview

Problem

Decoder bottlenecks in codec systems, particularly those using Run-Length and Entropy encoding techniques, lead to insufficient data output rates and loss of image data due to unpredictability in compressed data streams, limiting them to single decoding channels.

Innovation Solution

An encoder-decoder system that controls quantization during encoding to produce blocks with a predetermined number of zero coefficients, enabling parallel decoding and improving throughput by optimizing entropy and run-length decoding processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of substance

If Run-Length and Entropy encoding techniques are used for data compression, then data compression ratio is improved, but decoder throughput deteriorates due to unpredictability in compressed data streams

Engineering Contradiction:

Improvedata compression ratioVSAvoiddecoder throughput

Core Design Contradiction:

Loss of substanceVSProductivity

Solution Approach 1:

The encoder performs preliminary quantization control to shape the compressed data stream before entropy encoding. By pre-determining the number of zero coefficients in each block and controlling the run-length patterns, the encoder prepares the data in advance to enable predictable and efficient decoder operation, resolving the throughput problem while maintaining compression efficiency

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention changes the quantization parameter control strategy by explicitly controlling the number of zero coefficients (k values) in each block. This parameter control transforms the unpredictable compressed stream into one with predictable run-length patterns, allowing the decoder to optimize its processing speed and throughput while maintaining good compression ratios

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If single decoding channel is used to simplify decoder architecture, then device complexity is reduced, but decoder performance deteriorates due to bottlenecking

Engineering Contradiction:

Improvedecoder architecture complexityVSAvoiddecoder output rate

Core Design Contradiction:

Device complexityVSProductivity

Solution Approach 1:

The invention segments the decoding process by dividing the compressed data stream into multiple independent blocks, each with controlled run-length patterns. This segmentation allows the decoder to process multiple blocks in parallel using multiple decoding channels, increasing throughput while keeping each individual channel relatively simple in structure

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The invention introduces dynamic parallel processing capability to the decoder architecture. By controlling the run-length patterns in the compressed stream, the system can dynamically activate multiple decoding channels based on the data characteristics, enabling the decoder to adapt its processing capacity to match the input data requirements and achieve higher output rates

Inventive Principle:

Principle #15Dynamics

3Loss of substance

If quantization level is increased to improve compression, then data compression is improved, but decoder bottlenecking worsens due to increased processing complexity

Engineering Contradiction:

Improvecompression efficiencyVSAvoiddecoder processing speed

Core Design Contradiction:

Loss of substanceVSProductivity

Solution Approach 1:

The encoder performs preliminary quantization control to shape the compressed data stream before entropy encoding. By pre-determining the number of zero coefficients in each block and controlling the run-length patterns, the encoder prepares the data in advance to enable predictable and efficient decoder operation, resolving the throughput problem while maintaining compression efficiency

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention changes the quantization parameter control strategy by explicitly controlling the number of zero coefficients (k values) in each block. This parameter control transforms the unpredictable compressed stream into one with predictable run-length patterns, allowing the decoder to optimize its processing speed and throughput while maintaining good compression ratios

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS9002122B2System and method for improving decoder performance using quantization control

Publication Date: 2015.04.07 OMNIVISION TECHNOLOGIES INC

AI summary

A codec includes an encoder having a quantization level generator that defines a quantization level specific to a block of values (e.g., transform coefficients), a quantizer that quantizes the block of transform coefficients according to the block-specific quantization level, a run-length encoder, and an entropy encoder. The quantization level is defined to result in at least a predetermined number (k) of quantized coefficients having a predetermined value. The amount of data compression by the encoder is proportional to (k). The codec also includes a decoder having entropy and run-length decoding sections whose throughputs are proportional to (k). The decoder takes advantage of this increased throughput by further decoding coefficients in parallel using a plurality of decoding channels. Methods for encoding and decoding data are also disclosed. The invention is well-suited to quantization, entropy, and/or run-length-based codecs, such as JPEG.