Temporal coherent image segmentation and its applications

We propose a trapped-ball method for image segmentation, which is fast, supports non-uniformly colored regions, and allows robust region segmentation even in the presence of imperfectly linked region edges. We also introduce two applications by using the trapped-ball image segmentation, for temporal coherent animations generation and editing. First, we present a system for vectorizing 2D raster format cartoon animations by the segmentation method and background construction. The output animations are visually flicker free, smaller in file size, and easy to edit. And then we present an automatic method for online video stream stylization, producing a temporally coherent output video stream, based on the trapped-ball segmentation and optical flow. Our system transforms video into an abstract style with large regions of constant color and highlighted bold edges.