Noise-robust color edge detector using gradient matrix and anisotropic Gaussian directional derivative matrix. (April 2016)

• Record Type: Journal Article
• Authors: Wang, Fu-Ping; Shui, Peng-Lang
• Is Part Of: Pattern recognition. Volume 52(2016:Apr.)
• Page Start: 346
• Page End: 357
• Publication Date: 2016-04
• Subjects: Color edge detection -- Anisotropic Gaussian directional derivative matrix -- Color edge strength map -- Color edge direction map -- Gradient matrix; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2015.11.001
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract: In this paper, a noise-robust color edge detector using gradient matrix and anisotropic Gaussian directional derivative (ANDD) matrix is proposed. In order to alleviate the conflict between high edge resolution and noise robustness in the color Canny detector where the isotropic Gaussian kernels and gradient matrix of the R, G, B components are used, the ANDDs of the three components of a color image are arranged into the ANDD matrix. From its singular value decomposition (SVD), the ANDD-based color edge strength map (CESM) is constructed and is relevant to the pixelwise optimal fusion of the ANDDs of the three components in the sense of color edge enhancement. The merits and defects of the ANDD-based CESM and gradient-based CESM are contrasted to show their complementarity in color edge detection. The two CESMs are fused to develop a new color edge detector. It is compared with the color Canny detector and two recent color edge detectors. The results show that the proposed detector attains better detection performance for noiseless and noisy color images corrupted by white Gaussian noise or impulsive noise of small percentage. Highlights: A color edge detector using gradient matrix and ANDD matrix is proposed. The SVD of the ANDD matrix is used to determine CESM and CESD of a color image. The new detector has both high edge resolution and noise-robustness. It attains better performance than the two existing color edge detectors.