Salient Object Detection Using Water Flow Approach and Image Boundary Contrast Map

Abstract

A salient object in an image or video is the object which looks attractive to the human beings. A salient object has more prominent feature as compared to its background. Interpretation of a salient object is an easier task for human being as compared to systems. Thus, we need different salient object detection algorithms, which are feed into the computer to segment the salient object from the image or video. Detecting a salient object is a crucial task for post processing applications such as object detection, object classification, image compression and so on. Global feature elucidates that salient object is obtained by identifying the element covering major part of larger area. Furthermore, the differentiating boundary between the salient object and background is inseparable due to radiometric intensity which is represented as a local feature. In the research work, a novel salient object detection method has been proposed as Water Flow Driven using Minimum Barrier Distance and Image Boundary Contrast Map. The experimental results and analysis state that Precision-Recall, Mean Absolute and the F-measure has been significantly compared and improved with the existing work.