Rotation and Scale-Invariant Texture Classification using Log- Polar and Ridgelet Transform

In Texture analysis, the rotation and scale invariant texture classification is one of the challenging problems. This paper provides a new method for rotation and scale invariant texture classification using log-polar and ridgelet transform. The log - polar transform is applied to unknown rotated or scaled or combined rotated and scaled image followed by ridgelet transform. The log-polar transform is used to eliminate the effect of rotation and scale changes. The ridgelet transform deals effectively with line singularities. Standard deviation, energy, entropy, which are taken from the sub bands of the log-polar, ridgelet transformed image and global standard deviation are the features used in this paper. The texture similarity measurement is accomplished by using minimum distance criterion.