Performance Analysis of Hyperspectral Supervised Target Detection Algorithms

Abstract

Hyperspectral detection of manmade or natural targets is an emerging area of research over past two decades. This paper focus on the target detection algorithms which utilizes structured background and their performance is analyzed using established statistical techniques. We have considered supervised algorithms such as Adaptive Matched Subspace Detector (AMSD), Orthogonal Subspace Projection (OSP) and Hybrid Structured Detector (HSD). We have used Automatic Target Generation Process (ATGP) to estimate the background endmembers. This paper presents a detailed analysis of structured target detection algorithms and the performance of the algorithms is evaluated using the real hyperspectral HYDICE Urban image.