Method of reference images in the processing of noisy images of quasi-periodic surfaces

An algorithm for the nonoptimal correlation processing of noisy remote-sensing images of quasi-periodic surfaces (e.g., geological formations, crop fields, and the sea surface) is developed. A domain is determined within which changes in the frequency of the input image relative to a reference image do not reduce the detection probability below a specified threshold. In addition, a condition is obtained for the aperture of the optical correlation processor which makes possible the nonoptimal processing of the input image using a single reference harmonic transparency.