News

The book “Interactive Segmentation Techniques: Algorithms and Performance Evaluation” was published in July, 2013, by Springer Singapore in the series of “SpringerBriefs in Electrical and Computer Engineering / SpringerBriefs in Signal Processing Series”. The authors are Jia He, a PhD student and research assistent at USC, Professor Chang-Su Kim from Korea University, and Professor C.-C. Jay Kuo. The book mainly focuses on interactive image segmentation. It covers the methods which make use of the image features such as colors, edges, contrast, structure, etc., and extract the foreground object information indicated by the users’ interactions, and then provide user-involved segmentation results until the results are acceptable to the users. They also discussed how the existing methods tried to capture users’ desires, improve the segmentation accuracy, accelerate the interaction-process loop, reduce the computation complexity and thus generate reasonable results. Different methodologies may have strengths on particular images. For a particular segmentation task, user can choose and develop a proper method according to their analysis in the book.[expand title=”Read More…” swaptitle=”See Less…”] “Once we worked on a project to convert the 2D video to 3D, we faced a problem to segment the foreground object out.” Jia says. “Since for different images, the foreground objects may have different definitions, we have to provide an interactive segmentation tool to that task. We did a lot of research work on the interactive image segmentation, and found that there had been so many methodologies on this task, and people were trying to provide a segmentation tool that is much easier to control and is more efficient to obtain user desired results. It is hard to say which method is the best, since this topic is still on the way of research, [...]