Automatic Noise Identification in Images using Statistical Features

Noises are unwanted information which will be normally present in an image. It should be removed in such a way that the important information present in an image is preserved. De-noising an image is very active research area in image processing. There are several algorithms for de-noise but each algorithm has its own assumptions, advantages and limitations. The method proposed in this paper uses a simple pattern classification for noise identification. The significant idea is to obtain the noise samples and to extract their statistical feature for identifying the noise type. Simple image filters are used to get the noise samples and noise identification is achieved by using few statistical features. The method is capable of accurately classifying the type of noise besides the type of images used.