Optical Correlators for Minelike Target Detection: Theory, Modeling, and Evaluation,
Abstract
Optical correlators were analyzed and simulated with emphasis on the evaluation of their performance in minelike target detection. Diffraction theory was used in analyzing correlator performance. A model of a correlator system using discrete optoelectronic devices, spatial light modulators, and detector arrays was formulated. Discretization and quantization effects on correlator performance were analyzed. It was shown that at least eight gray levels are needed for a satisfactory detection performance. Filter displacement in the Fourier plan of a few pixels was shown to cause a drastic reduction in the correlation peak and the detectability of the targets. Additive noise to the input image was considered in the simulation studies. A minimum signal-to-noise ratio of -2dB (low CSNR) is needed for target detection in the specific examples used. Limited resolution images were evaluated and results show small (<20 percent) degradation in the performance of the correlator. It has been shown that a target with 50 pixels or more can be detected if its CSNR is 1.5 or higher. This suggests the advantage of reducing computation time significantly by reducing the pixel resolution of the images. Thresholding was studied for the detection of minelike targets. The peak-to-average intensity of the correlation was found to be the best parameter for thresholding.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1996
- Accession Number
- ADA313099
Entities
People
- Girardeau L. Henderson
- Mustafa A. Abushagur