Abstract

This paper describes a computationally efficient method for boosting the performance of correlation filters. The correlation filter is augmented with an array of histograms and the associated affinity numbers, obtained from the training image set utilized in construction of the filter. In the operational phase, histograms of sensor images are examined only at the image neighborhoods where the correlation filter provides initial indications of a target occurrence. The presence of a target is affirmed at image locations that pass both the peak cross correlation and histogram tests. Results of numerous experiments demonstrate that reinforcement of the spatial correlation filter with the trainer histograms leads to more robust Fourier filters for target detection and classification.

© 2014 Optical Society of America

Distortion tolerant correlation filter design

Kaveh Heidary
Appl. Opt. 52(12) 2570-2576 (2013)

Filter-feature-based rotation-invariant joint Fourier transform correlator

Farid Ahmed and Mohammad A. Karim
Appl. Opt. 34(32) 7556-7560 (1995)

Distortion-invariant pattern recognition with Fourier-plane nonlinear filters

Bahram Javidi and Dean Painchaud
Appl. Opt. 35(2) 318-331 (1996)

Application of supergeneralized matched filters to target classification

Kaveh Heidary and H. John Caulfield
Appl. Opt. 44(1) 47-54 (2005)

Computationally efficient video restoration for Nyquist sampled imaging sensors combining an affine-motion-based temporal Kalman filter and adaptive Wiener filter

Michael Rucci, Russell C. Hardie, and Kenneth J. Barnard
Appl. Opt. 53(13) C1-C13 (2014)