Some Aspects of Gaussian-Hermite Moments in Image Analysis

The normalized orthogonal (orthonormal) Hermite moments, or Gaussian-Hermite moments, have a good performance to recover the original image, although with a limited number of moments. Hence, Gaussian-Hermite moments can be used to characterize images patterns due to its good properties. The paper also compares the noise sensibility between Gaussian-Hermite moments and geometric ones and gives experimental results in some applications. It is shown that Gaussian-Hermite moments are effective tools for image analysis.