In computer vision segmentation of an image is the division of a given (digital) image into contiguous regions. In current computer vision algorithms the similarity of image parts is usually defined in terms of color and texture. The goal to automatically segment images into semantically meaningful parts is very difficult to achieve.