There are more and more applications of non-photorealistic rendered images, sketches and drawings. Several techniques for generating such imagery are widely known. The stochastic painting-based painterly image (and video) generation presented herein is a multipurpose image rendering and representation method, suitable for many purposes: painterly rendering, storing, compression or indexing. It incorporates many new features like multiscale edge following, stroke-set optimizations, templates, color morphology, etc. We will demonstrate that the presented technique (called enhanced Stochastic Paintbrush Transformation or eSPT) is suitable for fast high quality painterly rendering, providing good lossless painted compression ratios and features that make it suitable for many applications.One of these we wish to emphasize is the suitability to code painted images in a way that does not introduce any coding artifacts (blockiness, ringings, etc.) but provides a compact form of representation that still retains the main property of a painting: that it is a painting after all.
|Number of pages||4|
|Journal||Proceedings - International Conference on Pattern Recognition|
|Publication status||Published - Dec 17 2004|
|Event||Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004 - Cambridge, United Kingdom|
Duration: Aug 23 2004 → Aug 26 2004
ASJC Scopus subject areas
- Computer Vision and Pattern Recognition