Feature Distribution Relaxation for Robust Pattern Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pattern identification methods are not robust against contrast variations caused by shading in image data and do not achieve high-speed processing simultaneously.
Innovation Solution
An information processing apparatus that extracts a feature distribution from input image data, relaxes localities of the feature distribution using a Gaussian filter, and uses sampling data to determine specific patterns, integrating results from weak classifiers to achieve robust and high-speed identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If pixel luminance difference method is used for high-speed pattern identification, then processing speed is improved, but robustness against contrast variations caused by shading deteriorates
Solution Approach 1:
The patent introduces an intermediary conversion process that transforms the original image data into a converted image data through a specific conversion formula. This intermediary representation preserves the essential pattern information while eliminating sensitivity to contrast variations and shading, thereby resolving the contradiction between high-speed processing and robustness.
Solution Approach 2:
The patent applies parameter changes by transforming the image data using a conversion formula that modifies the luminance values. This parameter transformation converts the original luminance information into a new representation that is invariant to contrast variations, enabling both high-speed processing and robust pattern identification.
2Reliability
If edge extraction is applied to improve robustness against contrast variations, then reliability is improved, but processing speed deteriorates due to coarse feature distribution
Solution Approach 1:
The patent replaces the traditional mechanical edge extraction process with a mathematical conversion operation. Instead of using complex edge detection algorithms that produce coarse features, the patent applies a direct luminance conversion formula that maintains fine-grained information while achieving contrast invariance, thus improving both robustness and speed.
3Productivity
If conventional pattern identification methods are used, then processing speed is maintained, but information volume from feature combinations is insufficient
Solution Approach 1:
The patent transforms the feature representation by applying a conversion operation that effectively adds a new dimension to the data. This conversion creates a transformed feature space where combinations of sampling points yield richer information while maintaining the computational efficiency required for high-speed processing.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method enhances pattern identification performance by improving robustness against variations and noise, increasing the information volume from feature combinations, and enabling high-speed processing.
Implementation Method 1
a second feature distribution map generation unit that generates a second feature distribution map by applying a conversion required to relax localities of the first feature distribution map to the first feature distribution map
Data Source
AI summary
An information processing apparatus comprises: a registration unit adapted to register information required to determine at least one specific pattern in an image; an input unit adapted to input image data; a first generation unit adapted to extract a predetermined feature distribution from the input image data, and to generate a first feature distribution map indicating the feature distribution; a second generation unit adapted to generate a second feature distribution map by applying a conversion required to relax localities of the first feature distribution map to the first feature distribution map; and a determination unit adapted to determine, using sampling data on the second feature distribution map and the registered information, which of the specific patterns the image data matches.