Froth image segmentation based on adaptive ray cluster

Abstract

There are a number of noise areas, blur and adhesive froths in flotation image, with which classical watershed has no corresponding mechanism to cope. The paper proposes a segmentation algorithm of adaptive ray cluster, seed region are extracted by visiting every pixel just once and then noise areas are removed out. Ray cluster connected with the area of a seed is symmetrically launched from geometric center of the seed region, and froth boundary is adaptively extracted and amended by the gray distribution of ray cluster. The algorithm resolves over segmentation and improves the accuracy of segmentation.