Edge enhancement for autocorrelation and chord function processing

Abstract

Chord functions are descriptors of two-dimensional objects and are attractive candidates for shape recognition purposes. Before performing chord function analysis of an object it is necessary to enhance the object's outline. This can be done in a simple manner optically by high-pass filtering. If done as a separate process this results in a cumbersome system involving two optical configurations. However, it has been found that the preliminary edge enhancement stage can be avoided in many cases provided that the object is lighter than its background. The resulting autocorrelations and chord functions are very similar to those obtained by preliminary edge enhancement. It is demonstrated that this effect is due to “clipping” of the low frequency part of the power spectrum.