A statistical approach to interactive image segmentation

Interactive segmentation aims to separate an object of interest from the rest of an image. This problem in computer vision is known to be hard, and very few fully automatic vision systems exist which have been shown to be accurate and robust under all sorts of challenging inputs. Most of the previous works require users to trace the whole boundary of the object. When the object has a complicated boundary, or the object is in a highly textured region, users have to put great effort into iteratively correcting the selection. To solve this problem, most researchers rely on computing cost function for the purpose of control on the boundary of the region. Minimizing cost function yields a large, sparse system of linear equations. This paper presents a new interactive algorithm by introducing new equations for image foreground background separation. In addition we propose Edge stopping function, a powerful tool for automatically controlling the update of seed colors with the boundaries of the image. Results show that the new method works effectively and provides an alternative computational algorithm for building interactive image editing tools.