Salient Pattern Detection via Integrating Local Contrast and Minimum Spanning tree

This paper presents an approach to distinguish a salient object from an image. In this paper, we have introduced a robust method to detect the salient object from an image. We have utilized local contrast and edge color dissimilarity features to identify the salient regions. Firstly, Local contrast is used which is highly capable to extract the foreground salient object from background based on local contrast with its neighboring regions while foreground and background are more similar. Secondly, edge color dissimilarity measure helps to locate the salient region from other regions of an image having more cluttered background images. However, combination of local contrast and edge color dissimilarity features works better in salient region detection from an image. Experimental results demonstrate that proposed approach presents a good result which has been tested on two datasets ECSSD and SOD.