Real-time multiple target centroid extractor

The extracting of multiple targets in a field of view (FOV) is one of the key problems of the acquisition, tracking, and pointing technique, where a large number of computation and complicated operations must be done. In this paper, a new real- time multiple target centroid extractor is presented. The basic idea is that by using hardware circuit performing image segmentation and horizontal neighborhood analysis, the interesting target pixels are extracted from the raw image, then a digital signal processor carries out the vertical neighborhood analysis of target pixels and calculation of multiple target parameters. An algorithm of target pixels neighborhood analysis is presented for both convex and concave target shapes. A prototype system has been built which can extract parameters of about 100 targets in each FOV, offering centroid and average intensity for each target in the FOV from video signal in real- time.