Accurate object outlining by combining space-variant smoothing and interactive editing with the thinning process

Outlining of objects for two-dimensional images is of great importance in computer graphics, multimedia design, and medical applications. It is however, difficult to get a clear boundary with the existing automatic outlining methods, when a two-dimensional image contains inaccurate information. Here, we present a new algorithm to solve the problem by combining space-variant smoothing and interactive editing with a thinning process. By performing space-variant smoothing, we are able to extract boundary outlines based on local gradient variation and global gray-scale information. By using interactive editing, we are able to pick up a competing path with the one initially produced. This edition is accomplished by manipulating the gradient field. The new algorithm provides an effective way to correct the potential mistakes caused by inaccurate information in original images, and to input desired boundary lines for some local areas where the original image does not contain enough information for the outlining process. As a result, our algorithm makes it possible to extract more accurate outlines of objects with simple and intuitive manipulations.