基于支持向量机的CBERS-02卫星影像信息提取

Information extraction from CBERS-02 remote sensing image using Support Vector Machine

CBERS卫星是由中国空间技术研究院与巴西空间研究院联合研制的地球资源遥感卫星,CBERS-02卫星数据总体质量比CBERS-01卫星有所提高,本文利用支持向量机方法对CBERS-02卫星影像信息进行提取。研究中首先用6S模式对影像进行大气校正,然后选择RBF为支持向量机方法的核函数,并用交叉验证方法得到影响RBF核函数的两个最佳参数值进行学习完成信息提取,最后将提取结果制作成矢量图。通过研究得出用大气校正后的数据进行信息提取分类精度有所提高;与最大似然法和最小距离法相比,支持向量机方法分类精度较高。通过将研究结果与ETM+影像进行比较得出,CBERS-02卫星影像精度能够满足应用需求并能代替TM/ETM+数据开展研究工作。

The CBERS Program was born from a partnership between Brazil and China in the space technical scientific segment. Data quality of CBERS -02 is evidently higher than CBERS -01. However, until now there are few researches based on the CBERS remote sensing images. This paper extracts information from CBERS -02 remote sensing image using Support Vector Machine. The remote sensing image is atmospherically corrected with 6S mode at first. Then the paper chose Radial Basis Function as the kernel function in support vector machine method and used two best parameters getting from cross - validation method to train the whole training set and compute out the classification result. The classification result was transformed to vector image finally. It can be concluded that the classification overall accuracy will be improved using atmospherically corrected data, and Support Vector Machine method has the higher overall accuracy comparing with the maximum likelihood and minimum distance classification method. By comparing the result with the ETM ＋ image, it suggested that the precision of CBERS -02 image can meet the application demand and TM/ETM ＋ images can be replaced by CBERS - 02 image in research of information extraction.