A new image co-segmentation method using saliency detection for surveillance image of coal miners

Abstract

In this paper, we propose a co-segmentation method using saliency detection. The input image is first over-segmented into super-pixels, in which, their similarities are measured by the Bhattacharyya coefficients. The proposed method uses the combination of detection results of different detection methods on different types of color space to produce the originating regions, in which optimized linear coefficient combination is exploited. Experiments are performed on different image databases and results comparable to that of some current state-of-the-art methods are provided.