An interactive region merging method

摘要

The paper presents a novel region merging method based on the interactive information from users. An image firstly is partitioned into homogeneous regions by using an initial segmentation and the regions will be labeled by taking an interactive scheme. In this scheme, the users only roughly specify the position and main features of the object and background, then any region will belong to labeled regions i.e. object or background or unlabelled regions. A similarity rule is used to guide the merging process with the help of the users' markers. And then the object of interest is extracted from the image. Experiment results show that the proposed method is efficient for us to extract the object from the complex background.