COMBINING COMPONENT FEATURES FROM MULTIPLE IMAGE FRAMES.

Abstract

One of the common operations in image analysis is component labeling and geometric feature extraction (CLFE) which computes basic geometric features such as area, perimeter, centroid and enclosing box from segmented objects in binary images. Very often, due to hardware and memory limitations, it is not possible to process the whole field of view at once. The problem addressed is how to perform CLEF effectively, given the constraints of limited hardware and memory. It is assumed that the full field of view consists of an array (mosaic) of image frames where each image frame can be processed by CLFE. The basic idea is to extract additional features from each image frame such that only original and additional features can be saved and processed to give the final result (set of original features) corresponding to the full field of view. The algorithm and its hardware/software implementations are described.