Gabor wavelet transform and application to problems in computer vision

B. S. Manjunath
Center for Information Processing Research
Department of Electrical and Computer Engineering
University of California, Santa Barbara, CA 93106 USA


An important goal of early vision is to transform the intensity data to a representation more suitable for further processing. Examples of such transformations include obtaining depth information from stereo images, shape from shading, image segmentation based on textural and intensity discontinuities, etc. It appears that one of the basic pre-processing steps in achieving these transformations is to detect oriented features at multiple scales. An approach that appears to be promising is the Gabor wavelet transformation of the intensity which results in such a multi-scale oriented feature representation. In this paper we review some recent work on the use of these wavelets in low level vision processing and illustrate some applications to problems such as boundary detection, feature detection and localization, and shape recognition.
[PDF] [BibTex]
B. S. Manjunath,
26th Asilomar Conference on Signals, Systems and Computers, pp. 796-800, Pacific Grove, CA, Oct. 1992.
Node ID: 214 , DB ID: 8 , VRLID: 14 , Lab: VRL , Target: Proceedings
Subject: [Image Texture] « Look up more