Image Recognition via Partial Image Segmentation and Group Matching

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image recognition methods are inefficient, particularly when dealing with a large number of classes, as they require extensive processing time due to the need for multiple image comparisons and dimensionality reduction does not significantly improve speed.

Innovation Solution

The method involves forming a dictionary of registration partial images and using machine learning to reduce the number of processing times by extracting and matching partial images, with a classification tree structure for efficient attribute discrimination.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If pattern matching method is used to recognize target objects, then recognition accuracy can be maintained, but recognition time becomes extremely long when the number of classes is large

Engineering Contradiction:
Improverecognition accuracyVSAvoidrecognition time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent divides the image recognition task into two stages: first segmenting the registration images into multiple groups based on visual similarity, then performing pattern matching only within each group. This segmentation reduces the overall matching complexity from O(N^2) to approximately O(N log N) by eliminating the need to compare every registration image with every test image.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary mechanism using visual similarity metrics and grouping structures as mediators between the registration images and test images. Instead of direct one-to-one matching, the system uses similarity-based grouping as an intermediate step to filter and organize the matching process, significantly reducing computational complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If the number of registration images is increased to improve recognition accuracy, then recognition accuracy improves, but processing time increases linearly

Engineering Contradiction:
Improverecognition accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the large set of registration images into multiple smaller groups based on visual similarity. This allows the system to maintain high recognition accuracy by preserving all registration images while reducing processing time through grouped matching, where images within each group are matched together rather than against all images.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the parameter of image grouping by introducing a visual similarity threshold and group structure. This parameter change transforms the matching process from individual image comparison to group-based comparison, fundamentally altering the computational complexity while maintaining accuracy.

Inventive Principle:
Principle #35Parameter changes

3Speed

If image collation is performed at very high speed, then processing time is reduced, but the complexity of position specifying task increases

Engineering Contradiction:
Improveimage processing speedVSAvoidposition specifying complexity
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The patent segments the image collation process into two phases: first identifying candidate regions through rapid similarity matching within groups, then performing precise position specifying only on the reduced set of candidates. This segmentation enables high-speed processing while maintaining position specifying accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by performing exhaustive pattern matching only within smaller image groups rather than across all images. This partial matching approach achieves high processing speed by reducing the scope of complex position specifying operations to only the necessary candidate regions.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2443612B1Image recognition method and image recognition apparatus
Publication Date: 2021.03.03 CANON KK
  • EP2443612B1 patent drawingFigure 1
  • EP2443612B1 patent drawingFigure 2A~2B
  • EP2443612B1 patent drawingFigure 3~4

AI summary

An image recognition apparatus is provided which comprises a first extracting means for extracting, from every registration image previously registered, a set of registration partial images of a predetermined size, and a second extracting means for extracting, from an input new image, a set of new partial images of a predetermined size. The apparatus further comprises a discriminating means for discriminating an attribute of the new partial image based on a rule formed by dividing the set of the registration partial images extracted by the first extracting means, and a collecting means for deriving a final recognition result of the new image by collecting discrimination results by the discriminating means at the time when the new partial images as elements of the set of the new partial images are input.