Segmentation and recognition in visual chromatic spaces

Abstract

The discriminatory additional information supplied by chromatic components of an image is useful for segmentation techniques that allow fast and efficient separation of the objects in a scene. We show a method that has two essential processes for separating and recognizing pans of a scene in the UVY visual chromatic space. In the learning process, each part of the scene is analyzed, extracting all its discriminatory chromatic features. For this purpose we use a clustering analysis, which divides the scene in several sections with similar chromatic features. This set of features constitutes the “learning bank.” The recognition process consists of recognizing, using the learning bank, each section obtained in clustering analysis. © 1990 Taylor & Francis Group, LLC.