Non-Gaussian Noise Detection by Machine Learning Algorithm for Multispectral Satellite Images

Satellite images are in vital role in case of studying and analyzing the large earth surface. The multiple satellite sensors are capturing the earth information with different resolution based on various applications. There are various reasons for adding the noises into images, with the effect of this, the satellite pixels are giving the misclassification of results. The removing of these pixels is needful for improving the efficiency of classification results. The identification of noise and removing can be done by using this dimensional reduction machine learning algorithm. The standard filters are not given sophisticated results for giving optimum results in case of global and local images. Overcome this limitation by using this proposed algorithm. In this method, the principal component analysis method and windowing filters are implemented for removing non-Gaussian noise. The PSNR and SSIM values are improved compared to other standard methods.