2017 2 nd International Conference on Computer Science and Technology (CST 2017) ISBN: 978-1-60595-461-5 Directional 3D Peer-Group Filter for Color Image with Random Impulse Noise Ling-Yuan HSU 1,a* , Hsien-Hsin CHOU 2 and Tung-Tsun LEE 2 1 St. Mary's Junior College of Medicine, Nursing and Management/Department of Information Management, I-Lan, Taiwan 2 National I-Lan University/Department of Electronic Engineering, I-Lan, Taiwan a HsuLingYuan@gmail. com *Corresponding author Keywords: Peer group, Color image, 3D directional weighted mean Abstract. Peer-group filtering is the most basic approach to image denoising in the spatial domain; however, the effectiveness of this method degrades rapidly with an increase in the amount of noise. In this paper, we propose a novel 3D (3 dimensions) directional peer-group filter (referred to as 3DPGF) for the restoration of color images through the removal of random impulse noise. As with other switching methods, 3DPGF proceeds through two steps. Noise detection stage is first performed using a directional peer-group method, whereupon a 3D peer-group weighted-mean technique is used to remove the noise. Simulation results demonstrate that 3DPGF is able to enhance the accuracy in the identification and removal of noise. Introduction Impulse noise is usually divided into two types: salt and pepper noise (S&P) and random noise. S&P noise is a fixed extreme value with values of either 0 or 255 in the red, green and blue channels of color images. In contrast, the intensity of random impulse noise is uniformly distributed across the entire luminance range [0, 255] in all three channels of a color image. Random noise and S&P noise are both produced via the mutation of a single pixel, and can seriously degrade the quality of images through the loss of information related to image detail. Over the last decade, several filtering algorithms have been proposed to overcome the problem of random impulse noise in color images. The vector median filter (VMF)[1] is one of the most effective with regard to noise suppression and computing efficiency. Unfortunately, when the noise ratio is high, some edges and other details can be destroyed by a traditional vector median filter. Various modifications to median filters have been proposed over the last three decades to deal with this problem. Trahanias and Venetsanopoulos [2] proposed a basic vector directional filter (BVDF) and Karakos and Trahanias [3] presented a directional-distance filter (DDF). BVDF and DDF enable the separate processing of vector-valued signals via directional processing in order to maintain image sharpness. This approach has proven successful in dealing with noise of high density. Most classical median type vector-based filters introduce too much smoothing, regardless of whether the pixels in the image are corrupted by impulse noise or not. Other well-known vector filters, such as adaptive vector median filter (AVMF)[4], robust switching VMF (RSVMF)[5], edge detection based switching VMF 171