A comparative analysis of filtering techniques on application in image denoising

Noise is an inevitable part of real world images, so that efficient denoising is highly necessary, that leads to improve the performance. This makes the task of image denoising a great challenge for researchers. Denoising is a pre-processing task for the problems like identification, segmentation and classification. This paper deals with the filtering techniques using discrete cosine transform (DCT) which is a transform based filtering, median filtering, and bilateral filtering. The objective is to compare these types of filter for suitability of image denoising. Hence initially it is used for DCT based image denoising next to it the median filtering is applied and finally the bilateral filtering is applied and shown its efficacy. We found that for patch based processing, DCT based denoising and median filtering work well but for long dynamic range it requires the kernel based processing which satisfies the bilateral filtering approach. The comparison has been made based on visualization and measured the parameters of peak signal-to-noise ratio (PSNR) and SSIM.