【摘要】 针对高光谱异常检测算法没有充分利用高光谱图像的空间信息,检测精度受到限制的问题,提出一种融合空谱信息的RX (Fusing Spatial and Spectral Reed-Xiaoli, FSSRX)异常检测算法来提高高光谱的异常检测精度。FSSRX算法利用EMAP (Extended Multi-attribute Profiles)方法提取出高光谱图像的空间特征,在空间特征上进行RX异常检测,计算空间特征中每个像素点的异常得分;直接对原始高光谱图像进行RX异常检测,计算在光谱特征中每个像素点的异常得分;将在空间特征和光谱特征中得到的异常得分进行有效融合,以提高检测精度。仿真结果显示,FSSRX算法能够有效提高检测精度,降低虚警率,与其他几种算法相比,检测性能更佳。更多还原

【Abstract】 To address the problem that the hyperspectral anomaly detection algorithm does not make full use of the spatial information of the hyperspectral image and the detection accuracy is limited, a FSSRX(Fusing Spatial and Spectral Reed-Xiaol) anomaly detection algorithm that fuses spatial and spectrum information is proposed to improve the accuracy of hyperspectral anomaly detection. In FSSRX algorithm,the spatial feature of hyperspectral images is firstly extracted by the EMAP(Extended Multi-attribute Profile) method and the abnormal score of each pixel in spatial features is then calculated with RX detector. Meanwhile, RX anomaly detection is carried out directly on the original hyperspectral image to calculate the abnormal score of each pixel in the spectral feature. The anomaly scores obtained in spatial and spectral features are effectively combined to improve the detection accuracy. Simulation results show that FSSRX algorithm can effectively improve the detection accuracy and reduce the false alarm rate.Compared with other algorithms, FSSRX algorithm can achieve better detection performance.更多还原