A comparative study of Fourier descriptors and Hu´s seven moment invariants for image recognition

The paper evaluates and compares the performance of Fourier descriptors and Hu´s seven moment invariants for recognizing images with different spatial resolutions. Both Fourier descriptors and Hu´s seven moment invariants have the preferred invariance property against image transformations, including scale change, translation and rotation. However, spatial resolution thresholds exist for both of them. In our experiment with the image recognition engine, for Fourier descriptors, with feature vectors composed by the first 10 elements of the series, the spatial resolution should not be less than 64×64 to achieve 100% recognition. For Hu´s seven moment invariants, the minimum spatial resolution is 128×128.