This project supports collaboration by Dr. Anil K. Jain, Department of Computer Science, Michigan state University, East Lansing, and Dr. Muhittin Gokmen, Department of Computer Science, Istanbul Technical University, Istanbul, Turkey. The research deals with the area of image representation, edge detection, and surface reconstruction using regularization theory. Regularization is a general framework to solve ill-posed early vision problems. The Lambda-Tau (L-T) representation of images or surfaces obtained by means of regularization provides additional advantages over the scale-space representation which is widely used to solve vision problems. The L-T representation, derived from a hybrid energy functional, samples an image in both scale-space and Sobolev space of continuous functions. Edges and surfaces represented in this space have different levels of smoothness and continuity, controlled by parameters L and T, respectively. The main focus of the research will be on the following topics: the analysis of the L-T representation, and its use in solving early vision problems; the development of a generalized edge detector which can encompass most of the existing edge operators under a unified framework; and reconstruction of surfaces represented in the L-T space. Scope: This project brings together two teams with excellent capabilities in research in the field, and with complementary expertise to collaborate in an important area. Dr. Jain's Pattern Recognition and Image Processing (PRIP) Laboratory at MSU has published extensively on topics dealing with range data, surface reconstruction, and 3-D object recognition. Dr. Gokmen's research group has substantial expertise in regularization theory, edge detection, surface reconstruction and object recognition. The collaboration is expected to be beneficial to the two groups. This project meets the objectives of the Division of International Programs.
