| Many applications require locating a particular object in a collection of digital images. Motivated in part by the increasing spread of visual databases, various content-based image searching techniques have been proposed. Typically, characteristics are subsequently compared for similarity against features of targeted objects.; We present a novel adaptive-sampling algorithm designed to increase inter-object discrimination and reduce feature-vector dimensionality. Our algorithm is applied to a Gabor-feature based multi-resolutional object detection and recognition scheme. In this context we study and analyze the detection and identification of unknown objects in a complex background. Iterative, off-line optimization methods are employed to reduce computational demands during the learning phase.; Due to the large dimensionality of the feature set and the need to execute a database query for all pixels in searched images, we have implemented a Vector-Approximation (VA) based query system. The VA method comprises of a recently proposed multidimensional indexing structure and search algorithms. Several design issues pertaining to this indexing method are presented and analyzed. |
