Skeleton-based morphing between sketching images

摘要

Image morphing techniques is to produce continuous evolution between two images. It has practical application in scientific visualization and animation in films and advertising industries. Most traditional methods require users to prescribe correspondence with pairs of features between source and target images manually. However, if they have no similar features at all, it would be very hard for users to define the correspondence. In this paper, we proposed a sketching images morphing method to reduce user's burden by defining correspondence automatically. We extract skeletons from source and target sketching images. Then segment and represent them as sets of curves. By computing their centroids and adjusting their orders, we can obtain their reasonable mapping relationship. The morphing procedure is implemented by interpolating skeleton curves and stroke widths. Experiments show that our algorithm is fast, stable and easy to implement. Smooth transitions can be produced automatically between sketching images.