Auto-Corner Detection Based on the Eigenvalues Product of Covariance Matrices over Multi-Regions of Support

TLDR: An auto-detection corner based on eigenvalues product of covariance matrices (ADEPCM) of boundary points over multi-region of support is presented, which considers that points corresponding to peaks of eigen values product graph are reported as corners, which avoids human judgment and curvature threshold settings.

Abstract: In this paper we present an auto-detection corner based on eigenvalues product of covariance matrices (ADEPCM) of boundary points over multi-region of support. The algorithm starts with extracting the contour of an object, and then computes the eigenvalues product of covariance matrices of this contour at various regions of support. Finally determine automatically peaks of the graph of eigenvalues product function. We consider that points corresponding to peaks of eigenvalues product graph are reported as corners, which avoids human judgment and curvature threshold settings. Experimental results show that the proposed method has more robustness for noise and various geometrical transform.