Object detection using active contour model with depth clue

Abstract

This paper proposes a new method that uses depth clue with the classical contour models for detecting salient objects in noisy images. Unfortunately, most of the proposed stopping functions, including gradient and polarity, fail to detect objects effectively in many circumstances. On the other hand, depth disparity information, if available, could provide better clues for object detection. Thanks to the newly available low cost depth sensors, such as Microsoft Kinect, the use of depth disparity for solving object detection becomes reality. The proposed method takes the advantage of the existing contour models by using the depth disparity clues, from Kinect sensor, instead of two-dimensional clues, in the model stopping function. The depth disparity is applied in the external energy function of the active contour model for detecting the object of interest in the image. The experiments, carried out on real images, have shown the success and effectiveness of our proposed method to detect salient objects. © 2013 Springer-Verlag.