A robust FLIR target detection employing an auto-convergent pulse coupled neural network

ABSTRACT

Automatic target detection (ATD) of a small target along with its true shape from highly cluttered forward-looking infrared (FLIR) imagery is crucial. FLIR imagery is low contrast in nature, which makes it difficult to discriminate the target from its immediate background. Here, pulse-coupled neural network (PCNN) is extended with auto-convergent criteria to provide an efficient ATD tool. The proposed auto-convergent PCNN (AC-PCNN) segments the target from its background in an adaptive manner to identify the target region when the target is camouflaged or contains higher visual clutter. Then, selection of region of interest followed by template matching is augmented to capture the accurate shape of a target in a real scenario. The outcomes of the proposed method are validated through well-known statistical methods and found superior performance over other conventional methods.