Rotation-invariant texture Image Classification using R-transform

摘要

In this paper, An descriptor of rotation-invariant texture Image Classification is proposed. The feature exacted by using the proposed method is rotation invariant and robust to the change of spatial scale and illumination. Rotation invariant is achieved by means of Rapid transformation (R-transform) to eliminate cyclic translation resulting from rotation variation on the local circle region, which rounds a centre pixel. Combination of several descriptors with different (N, R) parameters the spatial multiscale are obtained. Texture classification experiments were carried out on the Brodatz databases and promising results are obtained from those experiments.