Salient Region Detection Using Contrast-Based Saliency and Watershed Segmentation

Salient region detection is useful for many applications such as image segmentation, compression, image retrieval, object tracking, and machine vision systems. In this paper, an approach to detect salient regions in a visual scene using contrast-based saliency and watershed segmentation is presented. The approach allows salient objects to be detected and extracted for analysis while preserving the actual boundaries of the salient objects. The approach can be executed in parallel making it efficient for real time applications.