Department of Image Processing

Orthogonal moments are powerful tools in pattern recognition and image processing. This presentation focuses on the study of orthogonal Gaussian-Hermite moments and their applications in image processing.

Image reconstruction from Gaussian-Hermite moments is detailedly discussed. The influence of the scale parameter σ is analysed and an automatic σ selection is proposed as well.

The development of the invariants to Gaussian-Hermite moments is also proposed. The fact shows that the rotation invariants of Gaussian-Hermite moments have identical constructing forms to those of geometric moments. Moreover, their central moments have translation invariance.

Finally, some applications of Gaussian-Hermite moments and their invariants are given, including orientation estimation, template matching, image registration and mosaic image generation.