An efficient method to build early image description

Abstract

The elaboration of an early image description based on an adjacency graph of reliable regions is discussed. An optimal edge detector is used first to extract the contours of the analyzed scene. A topological-configuration-based look-up table is then used to determine the extremity points, and a closing-contours process is applied using gradient magnitude information. An efficient connected components labeling algorithm is used to segment the image into reliable regions. A region-based adjacency graph, where radiometric parameters and reliable polygonal approximation of region boundaries are associated with each node, is then created and used as an early image description of the analyzed image.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>