Images are electronic snapshots of a scene or scanned from documents, photographs, manuscripts, printed texts, and artwork. Images when digitized are sampled and mapped as a grid of dots or picture elements (pixels). Different types of noises get introduced in the image, out of which speckle noise removal is considered fairly difficult. A lot of filters were created which basically worked on enhancing the quality parameters of the image, like Coefficient Of Correlation, Peak signal-to-noise ratio and Equivalent Number of Looks. Out of these models, Wavelet denoising, Lee, Frost and Kuan filters were quite famous. However, none were able to preserve the original image quality, as they required either removal of edges or over-smoothing for denoising the image. The objective of this paper is to work on the drawbacks of earlier models, and propose a modern approach to de-speckling an image by creating a hybrid filter which uses two different filtering techniques: wavelet denoising and anisotropic diffusion filter, to de-speckle the image without considerable removal of edges or causing over-smoothing.

Digital image despeckling wavelet transform anisotropic filter