改进型ICA和SVM相结合的火山灰云遥感检测

Remote Sensing Detection of Volcanic Ash Cloud Using Improved Independent Component Analysis and Support Vector Machine Algorithm

针对独立分量分析(ICA)模型在火山灰云遥感检测中的不足,提出了一种改进型ICA即变分贝叶斯ICA(VBICA)和支持向量机(SVM)相结合的火山灰云遥感检测算法,实现了火山灰云信息的近似分离。实验结果表明,所提算法能够从中分辨率成像光谱仪(MODIS)遥感图像中检测出火山灰云目标信息,且总检测精度和Kappa系数分别达到了88.4%和0.801 1,取得了较好的检测效果。

For the deficiencies of Independent Component Analysis( ICA) model in volcanic ash cloud remote sensing detection,a remote sensing detection algorithm is proposed based on improved ICA( namely Variational Bayesian ICA,VBICA) and Support Vector Machine( SVM) to realize the approximate separation of volcanic ash cloud information. Test results show that the proposed method can detect the volcanic ash cloud information from the Moderate Resolution Imaging Spectradiometer( MODIS) remote sensing image,and the total detection accuracy and Kappa coefficient reaches 88. 4% and 0. 801 1 respectively. The detection result is satisfying.