A Hyperspectral Imagery Anomaly Detection Algorithm Based on Gauss-Markov Model

Anomaly detection is an important fore-processing part in the hyperspectral imagery analysis chain because it can reduce the huge amount of raw data. In the conventional hyperspectral anomaly detection algorithm, the spatial correlation of the background clutters is often neglected. Moreover, the computational costs render the algorithm ineffective without significant data amount reduction. In this paper, an improved anomaly algorithm is proposed, assuming that the background clutter in the hyperspectral imagery is a three-dimensional Gauss-Markov random field. That is, each interested target may be considered with its contiguous regions during detection. The further anomaly detection algorithm is realized by constructing detection operator based on Gauss-Markov estimation parameters in hyperspectral imagery. Simulation results show that the proposed anomaly detection method based on Gauss-Markov model is more effective than the popular detection algorithm in hyperspectral remote sensing imagery.