An experimental comparison of superpixels detection methods for contour detection

Xuan-Yin Wang,Chang-Wei Wu,Ke Xiang,Sen-Wei Xiang,Wen Chen

DOI: https://doi.org/10.1007/s00138-018-0927-x

2018-01-01

Abstract:Recently, many superpixels detection methods have been proposed and used in various applications. We are interested in which method is more suitable for the application of contour detection. In this paper, superpixels are evaluated on BSDS500 dataset in two different aspects. On the one hand, contours are directly provided by the boundaries of superpixels and experiments show that better results could be achieved by the superpixels with irregular shapes than those with regular shapes and similar sizes. On the other hand, contours are further detected from those candidate positions which are confirmed by the boundaries of superpixels through the operation of dilation. In this situation, experiments show that competitive results could also be achieved by some superpixels with regular shapes and similar sizes. Besides, we propose a superpixels detection method called watershed-based graph (WG), by which superpixels with irregular shapes could be produced. Firstly, a graph is constructed from an over-segmented map which is achieved through a watershed algorithm. Then, to get the desired superpixels, the graph is segmented by merging neighbor segments in an order of decreasing similarity. Experiments show that higher efficiency could be achieved by WG with a moderate worse contour quality than its original graph-based method.