Ameliorating the Saliency of Objects in digital Images

Abstract

Visual saliency is the capability of a revelation system of human or machine to choose a sub part of information for further processing. The mechanism describes here serves as a filter to pick only the interesting information related to given errands and the tasks while ignoring inappropriate information. This work uses Random Forest to know the similarity between the image patches, apply active contour model to get the approximate contour and do dynamic thresholding segmentation .The results we get consists of many small artifacts, so to remove the low level details and to obtain the more smoothness we apply gradient minimization technique.