Computational aspects of exponent-Fourier moments

In this research, some computational issues of the orthogonal Exponent-Fourier moments are analyzed. A parallelized computing methodology based on the k × k numerical scheme and matrix operations has been proposed and utilized to improve both accuracy and efficiency of Exponent-Fourier moments computing. To demonstrate the improvement of our new computing methodology, the image reconstructions on an image sized at 1024 × 1024 are conducted from higher orders, up to 1000, of Exponent-Fourier moments with satisfied performances. By performing the image reconstructions from partial orders of Exponent-Fourier moments, we have discovered that the radial orders n and Fourier orders m of the Exponent-Fourier moments preserve the image information in different orientations. Finally, the centrally symmetric properties of Exponent-Fourier moments are discussed.