Abstract

Recently, Bracewell introduced a new transform called the Hartley transform (HT). This transform has the distinct advantage that HT of a real signal is also real. Thus, use of HT can potentially overcome some of the problems associated with Fourier transform (FT)-based methods which require the detection of complex wavefronts. Recently, Vijaya Kumar¹ proposed a new algorithm to compute image moments (useful for pattern recognition) from the Hartley transform intensity. We investigate the accuracy of moments computed using this algorithm when the algorithm is subjected to realistic impairments. The impairments considered include image plane noise, HT plane noise, sampling interval in the HT plane, dynamic range of the HT plane detectors, alignment requirements, and limited number of HT plane detectors.

© 1987 Optical Society of America