Scattered mines are one of the most important weapons in modern wars.Hyperspectral images are suitable for scattered mines detection given their abundant spatial and spectral information.This paper improves the traditional anomaly detection algorithm by effectively employing the distribution,size,and density knowledge of scattered mines,and proposes a knowledge-based detection algorithm for scattered mines.Compared to the traditional RXD algorithm,the pro-posed method improves the accuracy by about 66％while ensuring a higher recall rate,and the false alarm rate is more than 70％lower than the traditional algorithm.
维普
万方数据