A Smart Design of Pre-processing Classifier for Impulse Noises on Digital Images

摘要

Even though many kinds of filters algorithms have been developed for cancelling impulse noise from digital images in the past few decades, the issue of image restoration and reconstruction is still regarded for many researchers nowadays. In this study, a smart design of pre-processing classifier is used to categorize several distribution types which were identified as noise by noise detection. This smart design can be used for extracting useful local information from the corrupted image and hopefully results more image details to preserve for image filter. The pre-processing classifier goes through four-phase detection procedures to determine the condition of central pixel of local image window by using the similarity between neighbouring pixels. Simulation results show that our algorithm has extraordinary effective and accurate ability for classifying the condition of noise type.