Contour Corner Detection with Region Proportion for Overlapping Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing contour-based corner point detection methods in image processing have high computational complexity, making them inefficient for high-efficiency detection scenarios, especially when dealing with overlapping objects in industrial automation.
Innovation Solution
A method that determines corner points based on the region proportion of a predetermined region centered on the corner point, filtering out interference points through numerical comparison and utilizing location relationships with the object's center, and employing regions of interest to reduce the number of traversals and algorithm complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing contour-based corner point detection methods are used, then detection accuracy is improved, but computational complexity increases and detection efficiency decreases
Solution Approach 1:
The patent segments the corner point detection process into distinct phases: contour extraction, corner point candidate identification based on curvature, and filtering using region proportion. This segmentation allows each phase to focus on specific tasks, reducing overall algorithmic complexity while maintaining detection accuracy.
Solution Approach 2:
The patent introduces region proportion as a new parameter to filter corner point candidates. By calculating the proportion of the contour within a predetermined region around each candidate point and comparing it against a threshold, the algorithm efficiently eliminates false positives without requiring complex geometric calculations, thus reducing computational complexity while preserving accuracy.
2Measurement precision
If existing contour-based corner point detection methods are used, then detection accuracy is improved, but detection speed decreases
Solution Approach 1:
The patent performs preliminary filtering of corner point candidates using curvature-based identification before applying the more computationally intensive region proportion calculation. This preliminary action reduces the number of points that require detailed analysis, thereby improving detection speed while maintaining accuracy for the final selected corner points.
Solution Approach 2:
By introducing the region proportion parameter with a threshold comparison mechanism, the patent creates an efficient filtering step that quickly eliminates non-corner points. This parameter-based filtering approach is computationally lighter than traditional methods, significantly improving detection speed while preserving accuracy.
3Productivity
If region proportion filtering is applied to all corner points, then detection efficiency is improved, but computational overhead increases
Solution Approach 1:
The patent segments the processing workload by applying region proportion filtering only to corner point candidates that have already been identified through curvature analysis. This segmented approach ensures that the computationally intensive region proportion calculation is performed only on relevant candidates rather than all contour points, reducing overall computational energy consumption while maintaining detection efficiency.
Solution Approach 2:
The patent applies region proportion filtering selectively to corner point candidates rather than all points on the contour. This partial application of the filtering mechanism reduces computational overhead and energy consumption by focusing resources only on points that have a higher probability of being true corner points, as identified by the preliminary curvature analysis.
Data Source
Figure 1~2
Figure 3
Figure 4(a)~4(g)
AI summary
Embodiments of the present application disclose a corner point detection method and apparatus, which may improve the efficiency of corner point detection. The corner point detection method includes: obtaining a plurality of corner points of a target object in an image to be detected, where the plurality of corner points are corner points on a contour of a first object in the image to be detected, and the contour of the first object includes a contour of the target object; obtaining a region proportion corresponding to each of the plurality of corner points, where the region proportion is a proportion of a region of the first object in a predetermined region centered on the corner point; and determining a target corner point of the target object from the plurality of corner points based on the region proportion.