Tile Boundary Image Encoding via Selective Wavelet Coefficients

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image compression techniques, such as JPEG 2000, face challenges in maintaining tile continuity and controlling the amount of encoded data, leading to noise at tile boundaries and increased image degradation when trying to suppress noise.

Innovation Solution

An image encoding apparatus that divides data into tiles, performs wavelet transforms referencing neighboring pixel data, and encodes only the necessary transformation coefficient data for each tile, reducing the amount of encoded data while maintaining tile continuity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If encoding processing includes pixels for the neighboring tile to maintain continuity, then image degradation at tile boundaries is suppressed, but the amount of encoded data becomes large

Engineering Contradiction:

Improveimage quality at tile boundaryVSAvoidamount of encoded data

Core Design Contradiction:

Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent extracts and encodes only the specific transformation coefficient data from the neighboring tile that is necessary for decoding the pixel data of the encoding target tile, rather than encoding all transformation coefficient data. This selective extraction approach maintains tile boundary continuity while reducing the total amount of encoded data.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial action by performing wavelet transform on pixel data including neighboring tile pixels, but then selectively encoding only the portion of transformation coefficient data that is necessary for reconstructing the encoding target tile. This avoids the excessive action of encoding all neighboring tile data while still achieving continuity suppression.

Inventive Principle:

Principle #16Partial or excessive action

2Quantity of substance

If quantization step is increased to reduce encoded data amount, then encoded data amount decreases, but image degradation becomes larger

Engineering Contradiction:

Improveamount of encoded dataVSAvoidimage quality

Core Design Contradiction:

Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent changes the parameter of which transformation coefficient data is encoded by selecting only the necessary coefficients for decoding the encoding target tile pixels. This parameter change allows maintaining smaller quantization steps for the selected coefficients, thereby preserving image quality while reducing overall encoded data amount.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS10003821B2Image encoding apparatus and method of controlling same

Publication Date: 2018.06.19 CANON KK

AI summary

To suppress image degradation of a tile boundary while suppressing an increase of an amount of encoded data due to tile division. When performing a wavelet transform on a tile of interest, a wavelet transform unit performs a wavelet transform by referring to pixel data of the tile of interest and a neighboring tile that neighbors the tile of interest to obtain transformation coefficient data of a set resolution level. From transformation coefficient data obtained from the neighboring tile, the wavelet transform unit supplies a subsequent quantizing unit with transformation coefficient data of each resolution level that is closest to a boundary between the tile of interest and the neighboring tile, as well as all transformation coefficient data obtained from the tile of interest.