Abstract

Optical correlation is one of important operations for many applications, for example, pattern recognition, feature extraction, and so on. With the development of optical information processing, many techniques have been applied to the optical correlation to meet different requirements. The conventional correlation is based on the Fourier transform,^1,2 therefore, it is space-invariant. However, when one pays attention to the correct location of the object and does not care for whether it has been rotated at some angle. In this case, not only the space-variant but also rotation-invariant correlation is needed. In this presentation, we combine the fractional correlation with the method of the circular harmonics decomposition, and present a method to achieve the locating rotation-invariant correlation.

© 1998 Optical Society of America