利用高斯马尔可夫随机场模型描述像元的邻域相关性信息,并将这种邻域信息引入到局域异常探测器中,提出了一种顾及邻域信息的高光谱遥感影像局域异常目标探测算法。实验证明,该方法克服了传统异常探测方法仅仅利用光谱信息的不足,比经典的RX算法的探测效果更好,并且可以更有效地探测出大于一个像元的异常目标。 Using the Gaussian Markov random field model to describe the neighborhood correlation information of pixels and introducing this neighborhood information into the local anomaly detector, a local hyperspectral remote sensing image with regional information is considered Target detection algorithm. Experiments show that this method overcomes the shortcomings of the conventional anomaly detection methods using only the spectral information and is better than the classical detection algorithm of the RX algorithm and can detect abnormal targets more than one pixel more effectively.