论文部分内容阅读
鉴于仅依赖光谱特征或纹理特征的传统溢油检测算法的信息检测精度较低的问题,本文提出了一种新的光学遥感数据的谱纹海面溢油检测方法。谱是光学遥感数据的油膜敏感波段图像,纹是利用灰度共生矩阵计算获得的图像纹理特征,将这些特征相结合,引入支持向量机方法(Support Vector Machine,SVM),建立谱纹海面溢油检测模型。本文以2006年渤海溢油事故为例,利用中等分辨率成像光谱仪MODIS的光学遥感数据对溢油进行检测,MODIS的第2波段为油膜敏感波段,所以,第2波段图像即为选取的谱特征,经过对各个纹理特征的分析得到,均值、对比和相关3个特征量可作为溢油提取的纹理特征。检测结果的总体精度达91.23%。试验结果表明,将MODIS图像的光谱特征和纹理特征相结合,可有效地对渤海海洋油膜信息进行检测,并具有很强的抑制噪声能力。
In view of the low accuracy of information detection of traditional oil spill detection algorithms, which rely solely on spectral features or texture features, this paper proposes a new spectral oil spill detection method based on remote sensing data. The spectrum is the oil-sensitive wavelength band image of the optical remote sensing data. The grain is the image texture feature calculated by using the gray level co-occurrence matrix. Combining these features, the introduction of Support Vector Machine (SVM) Test model. Taking Bohai oil spill accident in 2006 as an example, the optical remote sensing data of Moderate Resolution Imaging Spectroradiometer MODIS was used to detect oil spill. The second wave band of MODIS was the oil film sensitive wave band. Therefore, the second wave band image was the selected spectral characteristic After analyzing the texture features, the mean value, the contrast value and the related three feature values can be used as texture features for oil spill extraction. The overall accuracy of test results reached 91.23%. The experimental results show that combining the spectral features and texture features of MODIS images can effectively detect the marine oil film information in the Bohai Sea and has strong noise suppression ability.