Towards a novel image denoising method with edge-preserving sparse representation based on laplacian of B-spline edge-detection

• DOI: 10.1007/s11042-015-3097-0
• Publication Date: 2015-11-26T07:25:06Z
• Authors: Dongqing Xu; Xiuyou Wang; Gang Sun; Huaimin Li
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

To address the edge structure preservation problem in sparse representation image denoising, a Laplacian of B-spline (LOBS) edge detection operator was brought out, which solves the problem of singleness of existing edge-detection operators under noise environment and lack of robustness to noise to some extent. Based on LOBS operator, a novel sparse-based edge preservation image denoising method (ESRIDM) was proposed. It determines edge region by computing gradient with LOBS operator. The non-edge region was denoised normally, while noise in edge region can be filtered by setting appropriate threshold. Simulation experiment compared with Laplacian-of-Gaussian (LOG) operator and Canny operator shows that the LOBS edge-detection operator has better robustness and lost less edge. Denoising experiments on general images and video monitoring images show that this novel method can achieve better denoising effect in subjective vision.