Saliency detection via multiple-morphological and superpixel based fast fuzzy C-mean clustering network

To model a human perception-based saliency detection algorithm in cluttered and noisy background images is a challenging problem in computer vision. Recently, many saliency detection algorithms have been proposed, which exploit the background information, or boundary priors of an image, to detect the salient object similar to the human attention system. These algorithms may not provide satisfying detection results for color images due to the assimilation of local spatial information. In this paper, we propose an unsupervised saliency detection technique, which uses multi-color space-based morphological gradient images. These gradient images contain different edge features, which are useful to obtain an accurate counter-based superpixel image containing both foreground and background clusters. A robust background measuring technique is implemented to remove background clusters, which describes the spatial information of an image cluster to image boundaries. This geometric clarification method effectively removes multiple low-level clues to produce a precise and uniform saliency map. These initially obtained saliency maps are fused using a multi-map fusion technique, and a compact saliency map prevails. The proposed algorithm is evaluated by executing different experiments on nine data sets. The results show that the proposed algorithm performs well for the detection of both single and multiple objects. The proposed algorithm is computationally efficient and provides better saliency detection results than state-of-the-art algorithms.