Perceptual grouping (or organization) can be defined as the ability to detect
organized structures or patterns in the presence of missing and noisy information. It has been
shown to be of fundamental importance in computer vision, providing reliable information to
higher level functions, such as object detection and recognition. In this talk, I will consider the
problem of grouping oriented segments in highly cluttered images and present an iterative,
multi-scale tensor voting methodology for segmenting natural shapes in cluttered backgrounds.
In this context, segments are represented as second-order tensors and communicate with each
other through a voting scheme that incorporates the Gestalt principles of visual perception.
The key idea is removing background segments conservatively on an iterative fashion, using
multi-scale analysis, and re-voting on the retained segments. We have performed extensive
experiments to evaluate the strengths and weaknesses of our approach using both synthetic
and real images from publicly available datasets. Our results and comparisons show improved
segmentation results, especially under severe background clutter.