Automatic noise identification in images using moments and neural network
Perumal Vasuki, Chintalapudi Likhitha Bhavana, S. Mohamed Mansoor Roomi, E. Lakshmi Deebikaa · 2012
Identifying noise from the original image is still a challenging research in image processing and is essential in order to counter the effects of unnecessary filtering process. Noise gets added to an image during image capture, transmission, or processing and degrades the performance of any image processing algorithms. Prior to de-noising step, the image should be tested for the identification of noise. Though Several approaches have been introduced in literature earlier for noise identification, each has its own assumption, advantages are not generic. This paper proposes a novel method based on statistical features with neural network classifier to identify the different types of noises such as Additive white Gaussian Noise, Salt & pepper Noise, Speckle Noise in the image without the human intervention. Extensive simulations on variety of images show that the proposed method effectively identifies the noise in a given image.