Directional line detectors in correlated noisy environments.

Abstract

In this research, we present a methodology for extracting very narrow lines in correlated noisy environments. The approach is a generalization of the analysis of variance applied to the symmetrically balanced incomplete-blocks design. It encompasses many well-known algorithms when subjected to more restrictive conditions. The detector is robust and superior to the polynomial-approximation-based detector and the classical Prewitt detector. The procedure detects narrow lines embedded in nonuniform background without compromising resolution, and performs satisfactorily in severe corruptive noise. Extensive computer simulations demonstrate the practicality of the detector on real imaging environments.