Some Remarks on FCMLS with Its Application to Foreground Extraction

It is well known that fuzzy clustering and level set are two important tools for image segmentation. The former focuses on analyzing the statistical characteristics of image features, while the latter aims to acquire the good geometrical continuity of segmentation boundaries. Obviously, two kinds of methods may be complement to each other. Inspired by this idea, a new level set model integrated with fuzzy c-means (FCM) clustering FCMLS has been presented in our previous studies. Compared with FCM and original level set methods, some remarkable characteristics and better performance have been demonstrated. In this paper, we mainly concern with the foreground extraction problem, wherein some extended works are also reported including detailed analysis on the convergence of FCMLS, multiregional FCMLS. Our research results indicate that FCMLS has good performance and is very worthful for foreground image extraction.