2-dimensional object recognition for the ISAC system

Abstract

The intelligent soft arm control (ISAC) system is a voice-activated robotic aid for the physically disabled which involves the requirement for object recognition. The authors present a brief description of the ISAC system and describe work in 2-D object recognition and orientation. The research in object recognition involves the extraction of relevant information from digitized images. A binary image was processed by a nonrecursive segmentation algorithm to isolate each object. Next, two histograms were generated for each object found by the segmentation routine: a distance histogram about the object's center of mass and an orientation histogram. The distance histogram was used for identification of objects and the orientation histogram was used for finding their orientation. Additionally, a nonrecursive segmentation algorithm, a histogram-based recognition and orientation detection algorithm and a multivariate-discriminant-analysis-based recognition algorithm are described.<>