Wavelet Transform Unit for DRAM-Free Image Encoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Real-time image transmission systems require efficient image encoding without the use of DRAM buffers to reduce cost and complexity, as existing 2D wavelet image encoding algorithms are not feasible in DRAM-buffer-less scenarios, such as surveillance cameras with constant transmission rates.

Innovation Solution

An image encoding system employing a wavelet transform unit, scan unit, quantizer, and entropy coder, which performs 1D, dual-line-based, or four-line-based wavelet transforms, and adaptive quantization, using SRAM buffers to store data, allowing for efficient compression and transmission without a DRAM buffer.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If 2D wavelet image encoding algorithm is used, then compression quality is improved, but device complexity and memory requirements increase due to DRAM buffer requirements

Engineering Contradiction:
Improvecompression qualityVSAvoidsystem complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent segments the image processing into multiple passes, with each pass handling a specific sub-band or frequency component. This allows the complex 2D wavelet transform to be broken down into simpler 1D transforms that can be executed sequentially with minimal buffering, resolving the contradiction between compression quality and system complexity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary sorting and reordering of wavelet coefficients before the main encoding process. By pre-organizing the data in a favorable order, the subsequent encoding requires less memory buffering and processing complexity, thereby maintaining compression quality while reducing device complexity.

Inventive Principle:
Principle #10Preliminary action

2Manufacturing precision

If 2D wavelet image encoding algorithm is used, then compression quality is improved, but loss of time increases due to DRAM buffering requirements

Engineering Contradiction:
Improvecompression qualityVSAvoidencoding time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent implements a continuous encoding process where the wavelet transform and encoding operations overlap in time. By using pipelining and keeping the encoder continuously busy with pre-processed data, the system achieves high compression quality without the time loss associated with DRAM buffering and data transfer.

Inventive Principle:
Principle #20Continuity of useful action

3Productivity

If DRAM buffer is used, then encoding efficiency is improved, but device complexity and cost increase

Engineering Contradiction:
Improveencoding efficiencyVSAvoidmemory subsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent extracts and removes the DRAM buffer component from the system by implementing algorithms that operate with minimal or no external memory buffering. The encoding efficiency is maintained through careful data management and processing techniques that eliminate the need for large DRAM buffers, thereby reducing device complexity and cost.

Inventive Principle:
Principle #2Taking out (Extraction)

4Loss of information

If tile-based coding with large tile size is used, then compression ratio is improved, but device complexity increases due to DRAM buffer requirements

Engineering Contradiction:
Improvecompression ratioVSAvoidmemory subsystem complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent transitions from spatial-domain tiling to frequency-domain processing by applying wavelet transforms. This dimensional change allows the system to achieve high compression ratios through frequency-selective encoding without requiring large DRAM buffers for spatial tile management, thereby reducing device complexity while maintaining compression efficiency.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS9531915B2Image encoding system and method thereof
Publication Date: 2016.12.27 ASPEED TECH
  • US9531915B2 patent drawing
  • US9531915B2 patent drawing
  • US9531915B2 patent drawing

AI summary

An image encoding system is disclosed. The image encoding system comprises a wavelet transform unit and a processing circuit. The wavelet transform unit performs a multiple-line-based wavelet transform on plural consecutive component lines to generate a wavelet transformed image comprising wavelet coefficients of plural sub-bands. The processing circuit coupled to the wavelet transform unit for quantizing, scanning and encoding the wavelet coefficients to generate a compressed image. Here, a number of the plural consecutive component lines is a multiple of 2 and less than 5. Since the wavelet transform unit performs the multiple-line-based wavelet transform to reduce the storage amount and maintain a good compression quality, an image encoding system of the invention can use SRAM buffers instead of a DRAM buffer.