Error-diffusion binarization for joint transform correlators

Research output: Contribution to journalArticlepeer-review


It is shown that the error-diffusion method is advantageous in the production of a binarized power spectrum interference pattern in joint transform image correlator (JTC) configurations, leading to better definition of the correlation location. A modified error-diffusion method for JTC configurations described here is effective also for cases involving noisy correlation planes. A computer-simulation comparison of correlation-peak quality, also in the presence of additive whitenoise, was carried out by using several performance criteria. The classical continuous JTC was compared with a binarized JTC, whereby the binarization method for a normalized and nonlinearly scaledFourier-transform interference intensity pattern was based on either error-diffusion or hard-cliping techniques that used a constant threshold value. The suggested operationsof normalization and nonlinear scaling are essentially equivalent to truncation and normalization. The error-diffusion binary JTC yields autocorrelation characteristics that are superior to those of the hard-clipping binary JTC over the whole nonlinear scaling range of the Fourier-transform interference intensity for all noise levels considered. The binary JTC, binarized in either the error-diffusion or the hard-clipping binarization method, is superior to the classical continuous JTC over a certain nonlinear scaling range and up to a specific level of additive-input white noise. In addition, the binarized cross-correlation JTC selectivity is generally better than that of the continuous JTC, especially when highlevels of additive-input white noise are present. Preliminary experimental results, which can beobtained in real time by using a single spatial light-modulator, are presented.

Original languageEnglish
Pages (from-to)707-714
Number of pages8
JournalApplied Optics
Issue number5
StatePublished - 10 Feb 1993


Dive into the research topics of 'Error-diffusion binarization for joint transform correlators'. Together they form a unique fingerprint.

Cite this