A robust texture descriptor using multifractal analysis with Gabor filter

Texture classification analysis and classification play an important role in the domain of content-based image retrieval, image segmentation, scene recognition and image/video analysis. This paper proposes a novel robust texture descriptor on variance in rotation, scale and illumination, which combines the dominant orientation analysis and multifractal analysis base on Gabor filter. The dominant orientations are extracted in corresponding Gaussian scales to handle the rotation variance, and then the scale and illumination invariant multifractal spectrum (MFS) is produced based on the multi-scale Gabor filters at the corresponding dominant orientation. The proposed approach is evaluated on Brodatz and Outex databases. The experiment results show that our method outperforms existing techniques under different condition variances.