RGB-D Salient Object Detection: A Review

Abstract

Salient object detection focuses on extracting attractive objects from the scene, which serves as a foundation of various vision tasks. Benefiting from the progress in acquisition devices, the depth cue is convenient to obtain, and is used in salient object detection in RGB-D images in combination with the color cue. In this chapter, we comprehensively review the advances in RGB-D salient object detection. We first introduce the task and key concepts in RGB-D salient object detection. Then, we briefly review the evolution of salient object detection technology, especially those for RGB images, since many RGB-D salient object detection methods derive from the existing RGB ones. Next, we present the typical RGB-D salient object detection methods, evaluate their performance on public datasets, and summarize their issues. Finally, we discuss some open problems and suggestions for future research.