M-band ridgelet transform based texture classification

The ridgelets overcome the shortcomings of wavelets and show great potential in texture classification. However, the ordinary rideglet transform inherits the weakness of the 2-band wavelet transform. That is, in the Radon domain, the wavelet transform decomposes a signal into channels that have the same bandwidth on a logarithmic scale. These characteristics are not suitable for analyzing the texture images, in which there are many edges (line singularities) that cause rich middle and high frequency components in the Radon domain. This paper will combine the M-band wavelet with the ridgelet and propose M-band ridgelet to overcome this disadvantage. The experimental results on two benchmark texture databases demonstrate the superior performance of the M-band ridgelet transform based texture classification.