Pyramid segmentation in 2-d and 3-d images using local optimization

A segmentation algorithm is presented which uses pyramid data structures as a means for achieving computational efficiency and for improving the quality of the segmentation. Results of experiments on synthetic and actual computed tomographic data are presented, and future enhancements to the algorithms are discussed. The motivation for using pyramids is to reduce computational costs through a coarse-to-fine segmentation strategy and to improve the quality of the segmentation.