When/Where: Thursday, May 1st, 4:15PM, 201 LOM

Speaker: Eitan Sharon, Division of Applied Mathematics, Brown University.

Title of Talk: A Hierarchical Image Description: Segmentation and Boundary
Detection using Multiscale Intensity Measurements

Abstract: Image segmentation is one of the most basic tasks for both computer and biological vision systems, and is a prerequisite for higher-level processes from motion detection to object recognition. Segmentation is difficult because objects may differ from their background by any of a variety of properties that can be observed in some, but often not all scales. A further complication is that coarse measurements, for detecting these properties, cannot be obtained by simple geometric averaging, because they would often average over properties of neighboring segments, making it difficult to identify the segments and to reliably detect their boundaries. I will present a novel method for segmentation, Segmentation by Weighted Aggregation (SWA). This algorithm efficiently detects segments that optimize a normalized-cut-like measure. The algorithm consists of an adaptive process in which pixels are recursively aggregated into increasingly large-scale aggregates of coherent properties: intensity, texture and boundary integrity. This allows us to detect regions separated by weak, yet consistent edges. The runtime complexity of the algorithm is linear in the number of pixels in the image, and the number of operations per pixel is very low. Numerous experimental results demonstrate a dramatic improvement over current state-of-the-art methods. In addition, the algorithm produces a novel description of the image as a hierarchy of segments. This hierarchical structure, which includes recursively- computed properties of segments at different scales (e.g., their "shape- moments"), is a natural basis for processes of learning and object recognition.

This is a joint work with Ronen Basri and Achi Brandt.