Image Processing Device Color Vector Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing devices using color images for pattern search often mistakenly detect regions due to ignoring the directional aspect of color distribution, leading to high normalized correlation values when relative color distributions are similar but not identical.
Innovation Solution
An image processing device that transforms the color variables of both the input and model images into single data arrays according to a predetermined rule, allowing for an absolute comparison of color variables rather than a relative one, thereby enhancing precision in identifying regions with high correlation values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If color difference is used as a scalar value in pattern search, then the pattern search process can be executed in color images, but the directional aspect of color distribution is ignored leading to false detections
Solution Approach 1:
The patent transforms the scalar color difference into a vector quantity by introducing directional information. The color difference is represented as a vector in color space that includes both magnitude and direction components, allowing the pattern search to consider the directional aspect of color distribution. This dimensional enhancement resolves the contradiction by maintaining color image processing capability while improving pattern matching accuracy through vector-based comparison.
Solution Approach 2:
The patent changes the parameter representation from scalar color difference to vector color difference. By transforming the single-value color difference metric into a multi-component vector that preserves directional information, the system maintains versatility in color image processing while eliminating false detections caused by ignoring color distribution direction. This parameter transformation directly addresses the accuracy issue without sacrificing adaptability.
2Device complexity
If normalized correlation value calculation ignores even concentration components, then calculation is simplified, but regions with similar relative color distribution but different absolute colors are mistakenly detected
Solution Approach 1:
The patent performs preliminary subtraction of even concentration components before calculating the normalized correlation value. By removing the uniform color component in advance, the calculation focuses only on the varying color distribution patterns. This preliminary action maintains computational simplicity while improving detection accuracy, as the correlation calculation now compares only the meaningful variable color patterns rather than being influenced by uniform background colors.
Solution Approach 2:
The patent extracts and separates the even concentration component from the color image data before performing pattern matching. By taking out the uniform color component that causes false correlations, the system maintains simplified calculation while improving reliability. The extraction isolates the relevant color distribution variations that truly indicate pattern matches, eliminating misleading correlations from uniform color regions.
3Adaptability or versatility
If relative color distribution comparison is used, then color variations are tolerated, but absolute color matching precision is lost causing false positives
Solution Approach 1:
The patent enhances the comparison from scalar to vector by incorporating directional information about color distribution. The vector representation preserves both the relative color relationships (providing variation tolerance) and the absolute color orientation (providing matching precision). This dimensional enhancement allows the system to distinguish between legitimate color variations and false positives by considering the direction of color differences in color space.
Data Source
AI summary
A determination target region includes a combination of a red gray image, a green gray image, and a blue gray image. A CPU part transforms a three gray value matrixes to a single first data array uniquely associated with a pixel position in the determination target region. The CPU part calculates a normalized correlation value with a single second data array transformed from a model image according to the same predetermined rule, and determines whether or not a match is found with the model image according to whether or not the normalized correlation value exceeds a predetermined threshold value.