Nonlinear Diffusion Filtering for Classifying Image

Abstract: Image processing is a signal processing for which input is an image output of an image� may be characteristics or parameter related to the image .Under this concept� the image classification which has both foreground and background feature of an image with edges, lines flow lines structures in foreground and inhibits smooth clutter in� background. In this paper the saliency driven nonlinear diffusion filtering algorithm is used and the image is classified using multiscale information fusion in original image, they are applied with diffusion process and finally mapped with saliency .Here the background image is considered as noise which improves image classification and finally they are removed using nonlinear diffusion filtering this process� makes the classification of images in an more accurate formation .At larger scales the background is filtered out and the foreground is preserved .Various experimental test has been conducted for image classification using multiscale space such as PASCAL2005 and oxford17 flower dataset with high classification rates

Keywords: PASCAL2005, multiscale space, nonlinear diffusion filtering