Shape Representation

Shape Representation in Computer Vision

Students: Ralf Juengling

Grant support: J. S. McDonnell Foundation

Good representations is the most important premise for significant progress in computer vision. Good representations should (1) facilitate abstraction or span several levels of abstraction, (2) allow representing sets of alternative hypotheses, and (3) facilitate integration of different kinds of evidence. Ralf Juengling's research focuses on representations of images and of two-dimensional shapes as a foundation for new algorithms in image segmentation, shape recognition and learning of shape models from examples.

Publications:

Juengling, R. and Mitchell, M. (2007). Combinatorial shape decomposition. In Proceedings of the Third International Symposium on Visual Computing (ISVC07). Springer (Lecture Notes in Computer Science).

Juengling, R., and Prasad, L. (2007). Parsing silhouettes without boundary curvature. In Proceedings of the 14th International Conference on Image Analysis and Processing, pages 665-670.