An effective denoising and enhancement technique for removal of random impulse noise in images

Abstract

Digital images are likely to suffer from impulse noise during the process of acquisition and subsequent transmission. In this brief a decision tree based method (DTBM) is presented which effectively removes random valued impulse noise from gray scale images. The pixels corrupted by noise are detected by an impulse detector and are reconstructed using an edge preserving image filter. The reconstructed pixels are written back in an adaptive manner so that they will be available for the next computation. The desirable feature of DTBM is that the design requires simple mathematical computations which results in a cost effective implementation of the system. The quantifiable evaluation and visual quality results of the presented system prove that the system's performance is good in both respects. Image enhancement is done on the denoised image using discrete cosine transform (DCT) which improves the visual quality of the image.