Classification of Images Based On Saliency Driven Non-Linear Diffusion Filtering

The saliency driven multiscale nonlinear diffusion filtering resulting scale space in general preserves or even enhances semantically important structures such as edges, lines, or flow-like structures in the foreground, inhabits and clear clutter in the background. The image is classified using multiscale information fusion based on the original image, the image at the final scale at which the diffusion process converges, and the image at a midscale. Our algorithm maintains the foreground features, which are important for image classification. The background image regions, whether considered as noise to the foreground, can be globally handled by fusing information from different scales. Index Terms – Nonlinear diffusion, image classification, multiscale information fusion.