Edge Aware Adaptive Filtering Method for Image Denoising

This paper presents an adaptive edge-preserving image denoising technique. The proposed approach adaptively adjusts the filtering mask size and mask content to perform image denoising. The mask size is determined depending on the distance from the nearest edge in the image. Different edge detectors (Prewitt, Sobel, Canny, Perona-Malik) were considered and validated. The filtering mask coefficients were selected using various kernels, such as mean and median. The filtering quality of the proposed solution in its several variants was compared with anisotropic diffusion and frequently used spatial filtering techniques like mean or median that operate globally. The qualitative and quantitative comparison of the proposed method is based on commonly known image quality measures as well as on edge preservation measures. The obtained results show the potential of the method: Quality Index up to 0.84 and edge preservation beta measure up to 98.49%.