Applying perceptual organization to the detection of man-made objects in non-urban scenes

Abstract

A new approach for the detection of large man-made objects in a non-urban area using a single monochrome image is presented. In this study, the man-made objects may be unspecified and the appearance of the objects is unpredictable. Prominent features that distinguish man-made objects from natural objects are identified. A computational framework of applying perceptual organization and using the prominent features is presented. Techniques are developed to group low level image features hierarchically into a region-of-interest (ROI) likely to enclose man-made objects or a substantial part of the man-made objects. These techniques include feature extraction, primitive structure formation, and segmentation. Some of these methods are novel and others present unique properties and advantages compared to previous related works. Experimental results are presented using real images that include several different man-made objects in complex backgrounds or a natural scene without man-made objects. It is shown that the located ROIs properly enclose the man-made objects in the scenes.