Survey on Image Denoising Methods

Noise removal is the vital need of every image processing tasks like segmentation, classification, object detection, etc. The SaP (Salt-and-Pepper) noise is projected by the highest and lowest intensity values. Many denoising methods are designed in the literature study to effectively remove the salt and pepper noise. This paper assesses image denoising to explore the positive and negative issues of eight benchmark methods. This work helps to the research people of image denoising for selecting the better method by considering various parameters.

I. Introduction

Noise in an image can be produced at the time of acquisition or during image transmission. The elimination of noise efficiently is the paramount preliminary process in image processing. The main objective is to retain the edges of images, textures and information of the fine-scale as much as possible from the noise corrupted original image. The past decade studies focus on the removal of various noises from images. Noise in an image can have serious effects on the quality of the digital image and there are several sources of noises are available. In Charge Coupled Device (CCD) imaging systems, the three main modules of noise are[1]: •

Photon Noise: It is the basic property of the quantum theory of light and it is a complicated source of noise because of the Poisson distribution nature of photon counting.

•

Readout Noise: The on-chip preamplifier is the major contribution of this noise. It is added uniformly to every image pels (picture elements). The utilization of high-end performance camera can significantly reduce the readout noise.

•

Dark Noise: It arises due to the statistical variation in the number of electrons which are thermally induced. It can be calculated and subtracted from the output of the image.