Image Compression Using Filter Information and Interpolation Coefficients

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image compression methods deteriorate image quality when transmitting high resolution images over limited bandwidth, as they transform high resolution images into low definition versions, leading to unnecessary bit transmission and quality loss.

Innovation Solution

An image compression apparatus that reduces image resolution and transmits filter information to prevent quality deterioration, using a filter set table to accumulate and reuse interpolation coefficients, clustering pixels into groups, and selecting representative or interest blocks to determine interpolation coefficients, thereby reducing bit transmission and enhancing image quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If high resolution image is transmitted using limited bandwidth, then image can be transmitted, but image quality deteriorates due to transformation to low definition image

Engineering Contradiction:
ImprovebandwidthVSAvoidimage quality
Core Design Contradiction:
Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent segments the image into multiple blocks and clusters them into groups, then determines interpolation filter coefficients for each group. This segmentation allows the system to transmit only essential filter information rather than complete high-resolution image data, resolving the contradiction between limited bandwidth and image quality preservation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary clustering of image blocks and determination of interpolation filter coefficients before transmission. By pre-processing the image to identify representative blocks and their corresponding filter coefficients, the system can transmit compressed data that maintains image quality when reconstructed at the receiving end.

Inventive Principle:
Principle #10Preliminary action

2Manufacturing precision

If upsampling interpolation filter coefficients are determined per pixel using 3x3 window clustering, then image quality can be maintained, but amount of bits increases due to unnecessary transmission

Engineering Contradiction:
Improveimage qualityVSAvoidamount of bits
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent merges adjacent pixels into larger blocks (e.g., 4x4, 8x8, 16x16) and clusters these blocks into groups that share common interpolation filter coefficients. This merging reduces the number of separate coefficient determinations needed, thereby reducing the total bits required for transmission while maintaining image quality through representative block selection.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent applies different block sizes and clustering strategies to different regions of the image based on their characteristics. By adapting the processing method to local image properties, the system optimizes the balance between image quality preservation and data compression, reducing unnecessary bit transmission in homogeneous regions while maintaining detail in important areas.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP2320655B1Apparatus and method of compressing image using filter information
Publication Date: 2018.01.10 SAMSUNG ELECTRONICS CO LTD
  • EP2320655B1 patent drawingFigure 1
  • EP2320655B1 patent drawingFigure 2
  • EP2320655B1 patent drawingFigure 3

AI summary

Provided is method and apparatus of compressing and restoring an image using filter information. The image compression apparatus may generate a reduced image by sampling an input image, determine filter information based on the input image and a decoded reduced image, and code the filter information and the decoded reduced image. Accordingly, a deterioration in an image quality may be prevented in transmitting the image using a limited bandwidth.