Spectral band independent recognition in a differential binary joint transform correlator

Abstract

Images of a target in a specific spectral band in general show no correlation with images of the same target in a different spectral band. Hence in a joint transform correlator (JTC) architecture, if the reference and input target are the images captured through a visible (e.g., charge-coupled device or CCD camera) and infrared (IR) detector, autocorrelation peaks are not obtained. This drawback has been overcome in this paper by the use of a CCD–IR fused image as the reference image. Daubechies wavelet transform, which produces the least root-mean-square (RMS) error in the fusion process in comparison to other wavelets, has been used for the purpose. A comparative analysis of the proposed idea has been carried out for the classical JTC (CJTC), binary JTC (BJTC) and differential binary JTC (DBJTC) algorithms. Since the DBJTC removes the dc completely and produces sharp correlation peaks compared to the other techniques, computer simulation and experimental results are shown for the proposed idea using DBJTC. The same fused reference image has also been used to identify multiple targets in a scene using DBJTC. Performance measures like correlation peak intensity (CPI), dc/ac and peak correlation energy (PCE) have been calculated as metrics of goodness for the proposed scheme.