Gradient Angle Line Detection for Varying Line Widths
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional line detection methods based on second derivatives fail to accurately detect lines when the line width exceeds a certain threshold, leading to incomplete or inaccurate detection.
Innovation Solution
A method utilizing the sum of gradient angle differences for neighborhood pixels around a center pixel to determine the presence of a line, employing a 3×3 window and specific threshold values to identify line candidates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a line detection method based on second derivative is used, then the detection process is simple, but the line detection accuracy deteriorates when the line width exceeds a specific width
Solution Approach 1:
The patent changes the detection parameter from second derivative magnitude to gradient angle difference sum. By calculating the sum of gradient angle differences for neighborhood pixels and comparing it with the center pixel, the method achieves accurate line detection for various widths without the limitations of second derivative-based approaches.
2Illumination intensity
If the line width is increased, then the line becomes more prominent in the image, but the conventional second derivative method fails to detect it accurately
Solution Approach 1:
The patent employs gradient angle difference as the detection parameter, which remains effective for thick lines. The sum of gradient angle differences for pixels surrounding the center pixel provides a reliable metric that detects lines accurately regardless of their width, solving the problem of missed detection for prominent thick lines.
3Measurement precision
If a 3×3 window is applied for gradient angle difference calculation, then the computational complexity increases, but the line detection accuracy for thick lines improves
Solution Approach 1:
The patent applies local quality by calculating gradient angle differences specifically for neighborhood pixels surrounding the center pixel within a 3×3 window. This localized approach focuses computational resources on the most relevant pixels for line detection, achieving high accuracy for thick lines while keeping the overall computational burden manageable through selective pixel processing.
Data Source
AI summary
Disclosed is a method for detecting a line in an image or a video. The method includes smoothing the image, selecting an area in which a line is to be detected, determining a center pixel in the selected area, applying a window centered on the center pixel, detecting a sum of gradient angle differences of pixels surrounding the center pixel, classifying the center pixel determined as the line candidate into a ridge or a valley, and eliminating a non-maximum pixel.