Perceptual Image Collation Using Human Visual Characteristics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image collation systems struggle to accurately determine whether two image data match, especially in cases where differences are not recognizable by humans, and are unable to handle halftone images effectively.
Innovation Solution
An information processing apparatus that converts image data into perceptual images based on human visual characteristics, using a two-stage collation process involving color-difference gradient and color-difference average calculations in a Region of Interest (ROI), allowing for ambiguous determinations similar to human visual recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional signal processing is used to determine image match/mismatch, then the determination process is simple and fast, but the system cannot make ambiguous determinations and cannot recognize differences imperceptible to humans
Solution Approach 1:
The patent introduces an intermediary component (the collation result determination unit) that mediates between the difference image generation and the final match/mismatch determination. This intermediary applies human visual characteristics to evaluate whether detected differences are perceptible, thereby resolving the contradiction between simple processing and accurate ambiguous determination.
Solution Approach 2:
The patent changes the evaluation parameters from simple pixel-level difference detection to human visual characteristic-based evaluation. By transforming the collation process to consider human perception parameters (such as visual acuity, contrast sensitivity), the system achieves accurate ambiguous determination while maintaining practical applicability.
2Reliability
If strict match determination is applied to halftone images, then printing quality standards are maintained, but the system incorrectly identifies imperceptible differences as errors
Solution Approach 1:
The patent changes the detection parameters from absolute pixel difference to human-perceptible difference. By incorporating human visual characteristics into the evaluation parameters, the system can distinguish between technically detectable differences and perceptually significant differences, thereby maintaining reliable quality assurance without false positives from imperceptible variations.
Solution Approach 2:
The patent applies different evaluation criteria to different regions and types of differences. Rather than uniformly treating all pixel differences as errors, the system evaluates differences based on their local context and perceptual significance, allowing strict quality control where needed while tolerating imperceptible variations elsewhere.
3Adaptability or versatility
If conventional image collation methods are used, then the process is straightforward, but the methods are unable to handle halftone images effectively
Solution Approach 1:
The patent performs preliminary processing to generate a difference image that specifically accounts for halftone characteristics before the final determination. By pre-processing the halftone images to highlight perceptually relevant differences and suppress irrelevant noise, the system gains halftone handling capability without requiring complete redesign of the collation system.
Data Source
AI summary
An information processing apparatus includes circuitry and a memory, the circuitry performing perceptual image conversion of converting two image data generated from one color image data respectively into perceptual images based on human visual characteristics, collation of determining whether the two perceptual images converted by the perceptual image conversion are same, and comparison result outputting of outputting the comparison result by the collation.