A New Method for Image Segmentation Based on Integration Technique
Huilan Luo, Wei Wang, Jing Li · 2010
There are some general procedures for image segmentation. One can apply different algorithms to create different clusterings of the data. Some clustering algorithms like k-means require initialization of parameters. Different initialization can lead to different data clusterings. In this paper, we explore the idea of evidence accumulation for combining the result of multiple clustering. Initially, data is decomposed into a large number of compact clusters; the K-means algorithm performs this decomposition, with several clusterings obtained by N random initializations of K-means. Taking the co-occurrences of pairs of patterns in the same cluster as means for their association, we deal with the image with the means. We compare our new method with the k-means. The experiment shows that our approach can achieve higher or comparable performance than the old method.