Using Perceptual Grouping to Detect Objects in Aerial Scenes

Mapping, cartography, photointerpretation and guidance are some applications that can directly and readily benefit from automated aerial scene analysis. We have successfully used perceptual organization ideas to analyze aerial scenes and to describe cultural features of interest, such as buildings. Perceptual organization refers to the ability of a visual system to quickly capture representations of structure and similarity among otherwise random elements, features, and patterns in the visual field. We describe some of the systems we have developed and applied to aerial images and present several examples. We also give a brief description of our current work, in particular the development of the concept of a grouping field to represent salinecy and help define the computational aspects of grouping operations. This paper is based on the work of many different people in our research group with more details of each part found in the referenced papers.