Wavelet Image Coding With Limited Levels And Differential Residual Encoding

Solution Overview

Problem

Existing image compression methods using wavelet transforms, such as Jpeg2000, face challenges in reducing the number of wavelet levels without introducing artifacts or discontinuities at tile edges during decompression.

Innovation Solution

A method that applies wavelet transformations to the entire image on a limited number of levels, ensuring a residual level of at least 2x2 dimensions, followed by differential coding using a quantification factor, to encode and decode images without discontinuities, by calculating the number of levels based on the image dimensions and using specific quantification factors for each level.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If the number of wavelet levels is reduced to decrease compression complexity, then computational resources are saved, but image quality deteriorates due to residual artifacts and discontinuities

Engineering Contradiction:

Improvecompression complexityVSAvoidimage quality

Core Design Contradiction:

Device complexityVSManufacturing precision

Solution Approach 1:

The image is divided into multiple tiles, and wavelet transforms are applied independently to each tile rather than to the entire image. This segmentation allows the use of fewer wavelet levels per tile while maintaining overall image quality, as each tile can be processed with limited levels without introducing discontinuities at tile boundaries due to the independent processing approach.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent changes the processing parameters by applying different quantization factors to different wavelet levels. Specifically, finer quantization is applied to higher levels while coarser quantization is applied to lower levels, optimizing the balance between compression efficiency and image quality when using a reduced number of wavelet levels.

Inventive Principle:

Principle #35Parameter changes

2Manufacturing precision

If wavelet transforms are applied to the entire image to maintain image continuity, then image quality is preserved, but the number of required wavelet levels increases computational complexity

Engineering Contradiction:

Improveimage qualityVSAvoidcomputational complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The image is divided into multiple tiles, and wavelet transforms are applied independently to each tile rather than to the entire image. This segmentation allows the use of fewer wavelet levels per tile while maintaining overall image quality, as each tile can be processed with limited levels without introducing discontinuities at tile boundaries due to the independent processing approach.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Instead of applying wavelet transforms to the entire image uniformly, the patent applies transforms to partial regions (tiles) with a limited number of levels, which is sufficient for each local region. This partial action approach reduces overall computational complexity while maintaining image quality through selective processing.

Inventive Principle:

Principle #16Partial or excessive action

3Productivity

If image division into tiles is used to reduce wavelet levels, then computational resources are saved, but discontinuities appear at tile edges

Engineering Contradiction:

Improvecompression efficiencyVSAvoidimage continuity

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The image is divided into multiple tiles, and wavelet transforms are applied independently to each tile rather than to the entire image. This segmentation allows the use of fewer wavelet levels per tile while maintaining overall image quality, as each tile can be processed with limited levels without introducing discontinuities at tile boundaries due to the independent processing approach.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentEP3056001B1Method for coding a matrix, in particular a matrix representative of a still image or of a video image, using a wavelet transform with a number of wavelet levels varying according to the image, and with different quantization factors for each wavelet level

Publication Date: 2021.08.18 COLIN

Figure 1~2
Figure 3~4

AI summary

The invention relates to an encoding method for encoding successive layers of an initial matrix into a compressed matrix and restoring it as a restored matrix, each cell of the initial matrix containing a respective initial numerical value; each cell of the compressed matrix containing a respective compressed numerical value corresponding to the respective initial numerical value; each cell of the restored matrix containing a respective restored numerical value corresponding to the respective initial numerical value; characterised in that to encode, one applies: - wavelet transforms applied to the entire matrix over a limited number of levels, such that a residual level remains with dimensions larger than or equal to 2x2; then, a differential encoding of the residual level, i.e., a compression of the residual level by obtaining differences with each of the values of the residual level.