Investigation of Various Noise and Denoise Methods In Satellite Images

Abstract

Variation in the brightness of an image is referred as its noise level. These variations affect the quality of the image. Noisy images manipulate the pixels in the image. Satellite images are prone to noise due to various reasons like environment conditions, capturing instrument, data transmission, defective sensors, faulty channels, etc. Denoising plays a major role as these images can be used in various applications .This paper proposes an investigation on types of noise in satellite images  like  Gaussian, Speckle, Salt & Pepper, and  Stripe and filters or algorithms to overcome these noises . Some parameters are discussed to give a clarity noise variation level like mean square error and peak signal to noise ratio. This paper focuses on Speckle and Gaussian  noise image  and comparative study  on filtering the speckle and Gaussian noise  using Logarithmic transformation and Nonlocal Means algorithm are executed .The denoised images can be of use in various fields of research. This paper also proposes a future work in building a noise analyzer in order to detect the type of noise in the images and suggest the best algorithm or filter to be used with the help of   Convolution Neural Network.