A New Translation,Scale,Rotation Invariant Analysis Approach to Object Recognition Based on Radon and Fourier-Mellin Transforms
Yao Li · 2008
In this paper, a new translation, scale, rotation invariant descriptor for gray images and twain images recognition is proposed, the proposed approach realizes translation invariant by coordinates transforms and utilizes Radon transforms to project the image onto projection space to convert the rotation to translation, and then applies Fourier-Mellin transforms to the result to covert the scale to amplitude variety, invariant features are extracted from the result. A k-nearest neighbors classifier is employed to classify objects. Theoretical and experimental results show the superiority of this approach compared with some recent methods, it can be applied to twain images and gray images with high robustness to additive noise. It is also shown that gray images can be classified with high accurateness, but the accurateness of classifying twain images is not high due to the errors introduced by Radon and Fourier-Mellin transforms.