Linear least-square modeling is a powerful modeling approach. Optimal piece-wise constant modeling is an important special case. In this paper we will develop a piece-wise constant modeling method that preserves the regional means and regional first absolute central moments, thus preserving important visual properties in images. The method will be used for two practical image applications: (1) image preprocessing techniques for better edge detection, and (2) dynamic range adjustment of images to meet display constraints. Our method yields better performance than existing standard techniques for those applications.
Proceedings Title: Proceedings of the International Conference on Imaging Science, Systems, and Technology : CISST '98
Conference Dates: July 6-9, 1998
Conference Location: Las Vegas, NV
Conference Title: International Conference on Imaging Science, Systems, and Technology
Pub Type: Conferences
dynamic range adjustment, edge detection, modeling, moment preservation