针对辽东湾海域的Hyperion高光谱遥感数据特点, 结合海面油膜光谱与Hyperion影像特征, 对该数据进行水陆分离与最小噪声分离(minimum noise fraction, MNF)变换处理, 在辽东湾海域MNF波段影像的2D散点图中, 海面油膜的出现会在其边缘形成一个异常散点区域, 可区分油膜与干扰信息,结合提取的海面油膜端元的MNF波谱, 通过混合调制匹配滤波(mixture tuned matched filtering, MTMF)技术, 成功地提取研究区海面油膜信息, 有效监测海面油膜信息, 为海洋环境监测提供新的技术手段。

Based on the Hyperion data of Liaodong Bay, with the spectral response and image characteristics of offshore oil slick analyzed, an effective method of extracting offshore oil slick image from Hyperion data was proposed. First, the ratio of 740nm to 690nm was used to extract water range from the Hyperion image of the research area; second, by transforming the water hyperspectral reflectance image with minimum noise fraction (MNF), the offshore oil slick information of every Hyperion band was converged into several MNF bands. Analysis on the 2 dimensional (2D) scatter plots of those MNF bands showed abnormal scatter plots would appear because of offshore oil slick, and those endmembers’MNF spectrum could be extracted by ENVI software’s 2D scatter plots tools. Taking those endmembers’ average MNF spectrum as standard offshore oil slick MNF spectrum, using the mixture tuned matched filtering method (MTMF) to filter the MNF images from 1 to 5 bands, and finally by evaluating the eigenvalue in matched filtering score image and infeasibility image value, the offshore oil slick could be success-fully extracted from the Hyperion image. This rapid detection method could be used to find offshore oil slick in hyperspectral images.