Feature extraction of trademark images using Geometric Invariant Moment and Zernike Moment - a comparison

The qualities of features extracted determine the successfulness of any image related applications. This is particularly true especially when the natures of images are occluded and consist of various shapes and design styles, such as trademark images. In this study Geometric Invariant Moment and Zernike Moment techniques are utilized to extract sets of features from trademark images. The results obtained are analysed and compared in terms of intraclass invariance to various perturbations. It is found that a set of features produced by Zernike Moment technique is more robust when compared to features produced by Geometric Invariant Moment.