Enhancing Gabor Wavelets Using Volumetric Fractal Dimension

摘要

Texture plays an important role on image analysis and computer vision. Local spatial variations of intensity and color indicate significant differences among several types of surfaces. One of the most widely adopted algorithms for texture analysis is the Gabor wavelets. This technique provides a multi-scale and multi-orientation representation of an image which is capable of characterizing different patterns of texture effectively. However, the texture descriptors used does not take full advantage of the richness of detail from the Gabor images generated in this process. In this paper, we propose a new method for extracting features of the Gabor wavelets space using volumetric fractal dimension. The results obtained in experimentation demonstrate that this method outperforms earlier proposed methods for Gabor space feature extraction and creates a more accurate and reliable method for texture analysis and classification.