Multilevel-DWT based image de-noising using feed forward artificial neural network

It is seen that during image acquisition, storage, retrieval or transmission, images get degraded due to presence of noise. With different varieties of noise and its extent, de-noising becomes challenging. Traditionally, a host of techniques have considered spatial, statistical and multiple domain approaches for de-noising. Yet, the scope always exist for exploring innovative means of performing de-noising for enhancing image quality. In the proposed work, we present an approach to de-noise images by combining the features of multilevel Discrete Wavelet Transform (DWT) and Feed Forward Artificial Neural Network (FF ANN). We apply our algorithm to de-noise the images corrupted by a kind of multiplicative noise known as speckle noise. The results show that the proposed method proves effective for a range of variations and is suitable for critical applications.