Adaptive Block-Based Key Point Location Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current image processing techniques for Visual Search and Augmented Reality face challenges in efficiently compressing location information of key points, leading to high bit-rates and reduced performance, especially at lower operating points, due to the sparsity of the histogram map matrix and the need for lossy compression methods that compromise retrieval accuracy.

Innovation Solution

An adaptive block-based compression method that differentiates between key points at the image center and periphery, using larger macro-blocks for the center and smaller ones for the periphery, combined with entropy coding and context modeling, to efficiently represent and compress the location information of key points, thereby improving compression rates without significant complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If uniform macro-blocks are used for compressing location information, then the compression process is simple, but the compression rate is insufficient and retrieval accuracy deteriorates

Engineering Contradiction:
Improvecompression process simplicityVSAvoidretrieval accuracy
Core Design Contradiction:
Ease of manufactureVSLoss of information

Solution Approach 1:

The patent applies different macro-block sizes to different regions of the image: larger macro-blocks (e.g., 16x16 or 32x32 pixels) for the image center where key points are densely distributed, and smaller macro-blocks (e.g., 8x8 or 4x4 pixels) for the periphery where key points are sparsely distributed. This local adaptation optimizes compression for each region's characteristics, improving overall compression rate while preserving retrieval accuracy.

Inventive Principle:
Principle #3Local quality

2Productivity

If larger macro-blocks are used for the image center, then the compression rate improves, but the complexity of the compression method increases

Engineering Contradiction:
Improvecompression rateVSAvoidcompression method complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the image into distinct regions (center and periphery) based on key point distribution characteristics. This segmentation allows the application of different compression strategies to each region, achieving high compression rates in the center while maintaining simplicity through clear regional boundaries and straightforward classification criteria.

Inventive Principle:
Principle #1Segmentation

3Loss of energy

If lossy compression is applied to reduce bit-rate, then the transmission efficiency improves, but the performance of visual search deteriorates

Engineering Contradiction:
Improvebit-rateVSAvoidvisual search performance
Core Design Contradiction:
Loss of energyVSReliability

Solution Approach 1:

The patent changes the parameter of macro-block size based on spatial location and key point density. By adapting the block size parameter to local characteristics, the method achieves efficient compression with lower bit-rates while preserving the essential location information needed for accurate visual search performance.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2801190B1Method for processing an image
Publication Date: 2018.08.15 HUAWEI TECH CO LTD
  • EP2801190B1 patent drawingFigure 1
  • EP2801190B1 patent drawingFigure 2
  • EP2801190B1 patent drawingFigure 3

AI summary

The invention relates to a method for processing an image, the method comprising: providing (101) a set of key points from the image; describing (103) location information of the set of key points in form of a binary matrix; and scanning (105) the binary matrix according to a predetermined order, thereby creating a new representation of the location information of the set of key points.