Rotation and Scale Invariant for Texture Analysis Based on Radon Transform and Wavelet Transform

A method of rotation and scale invariant for texture analysis based on radon transform and wavelet transform is proposed in this paper. First, after a texture is standardized, it is projected into Radon domain by Radon transform. Then the standard variance of the projecting data are computed at each projecting directions, thus all standard variances constitute the feature set of the texture. The principal direction is detected through the feature set. A feature curve is generated by standardizing the feature set based on the principal direction. In the end, the feature curve is decomposed by wavelet and the invariant features are extracted. By contrast with two other methods, our method exceeds in computation and accuracy.