论文部分内容阅读
针对高光谱图像空间分辨率较低导致异常检测虚警率较高的问题,提出了一种基于非下采样Contourlet变换融合的高光谱异常检测方法。首先对低空间分辨率的高光谱图像与高空间分辨率的全色图像进行NSCT变换;然后采用区域能量自适应加权的方法对低频子带系数进行融合,采用局部区域能量匹配的加权平均法与选择法相结合的方案对高频信息图像进行融合。最后运用SVDD算法对融合高光谱图像进行异常检测。利用真实AVIRIS图像进行实验,结果表明本文方法降低了异常检测的虚警率,具有更好的检测性能。
Aiming at the problem of high false alarm rate of anomaly detection due to the low spatial resolution of hyperspectral image, a hyperspectral anomaly detection method based on nonsubsampled Contourlet transform fusion is proposed. Firstly, the NSCT transform is performed on the low spatial resolution hyperspectral image and the high spatial resolution panchromatic image. Then, the regional energy adaptive weighting method is used to fuse the low frequency subband coefficients. The weighted average method of local area energy matching and The combination of selection method for high frequency information image fusion. Finally, the SVDD algorithm is used to detect the anomalies of the fused hyperspectral images. Experiments using real AVIRIS images show that the proposed method reduces the false alarm rate of anomaly detection and has better detection performance.