Contour Detection Variance Analysis for Low Reliability Feature Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing contour detection methods struggle to accurately identify portions with low reliability in automatic contour detection from images, particularly in cases where the detection target has features with varying brightness or color, leading to inconsistent results.
Innovation Solution
A contour detecting device and method that performs shape recognition multiple times, calculates variance for feature points, and determines points with a set value or greater variance as having low reliability, allowing for enhanced detection accuracy by displaying and highlighting these areas for user correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If shape recognition is performed multiple times to improve detection reliability, then the reliability of contour detection is improved, but the processing time and computational complexity increase
Solution Approach 1:
The system performs preliminary shape recognition multiple times before final contour determination, calculating variance for each feature point in advance. This preliminary multi-time recognition allows the system to identify unreliable feature points (those with high variance) and focus correction efforts only on those specific points, rather than reprocessing the entire contour detection pipeline.
Solution Approach 2:
The system applies different processing quality to different parts of the contour detection result. Feature points with low variance (high reliability) are accepted without further processing, while feature points with high variance (low reliability) are flagged for user correction or additional processing. This local differentiation optimizes processing time by avoiding redundant operations on already-reliable points.
2Measurement precision
If variance calculation is performed for all feature points to identify low reliability areas, then the accuracy of identifying unreliable regions is improved, but the computational complexity increases
Solution Approach 1:
The system replaces complex manual assessment of detection reliability with an automated statistical measure (variance calculation). Instead of requiring sophisticated algorithms or expert systems to judge which feature points are unreliable, the patent uses simple variance computation across multiple shape recognition results, which is computationally efficient and easily implementable.
3Manufacturing precision
If multiple shape recognition results are compared to determine low reliability feature points, then the accuracy of contour detection is improved, but the complexity of the detection process increases
Solution Approach 1:
The system creates multiple copies of the shape recognition process with slightly different initial conditions or parameters, then compares the results. By generating multiple detection results through replicated processing and comparing them via variance calculation, the system identifies consistent feature points (low variance) as reliable and inconsistent ones (high variance) as needing correction, simplifying the overall detection logic.
Data Source
AI summary
A contour detecting device, including: a display; and a processor, wherein the processor obtains a detection target image which is obtained by imaging a detection target, detects a contour of the detection target from the detection target image by performing shape recognition a plurality of times, the shape recognition recognizing a shape of the contour of the detection target, and causes the display to display a detection result of contour detection processing, and in the contour detection processing, for feature points forming contours obtained by performing the shape recognition the plurality of times, variance in the shape recognition that is performed the plurality of times is calculated for each corresponding feature points, and a feature point for which a value of the calculated variance is a set value or greater is determined to be a feature point which has a low reliability.