一种融合纹理特征与NDVI的随机森林海冰精细分类方法

海冰的精准分类对于掌握海冰生长发育状况，保障航海安全等具有重要意义。由于受数据源和分类方法等影响，使得海冰分类精度提高受限。本文面向高空间分辨率的光学遥感影像，提出了一种融合纹理特征和归一化差分植被指数（NDVI）的海冰精准分类方法，运用随机森林分类器构建海冰分类方法。以青岛胶州湾为实验区，高分二号（GF-2）为实验数据，进行了海冰类型提取，并与其他分类方法进行对比。结果显示:针对GF-2高分辨率光学遥感数据，融合纹理特征和NDVI的随机森林方法，相比于传统的随机森林、支持向量机、自动决策树和融合纹理特征的最大似然分类方法，总体分类精度分别提高13.70%、11.60%、19.22%、29.37%。Kappa系数分别提高0.16、0.13、0.22、0.44。相比于融合纹理特征和归一化水指数（NDWI）的随机森林方法，总体分类精度提高了9.67%，Kappa系数提高了0.09。这表明本文构建的海冰分类方法可有效提高海冰分类精度，为海冰的精确分类提供了一种有效的技术手段。

A fine classification method for sea ice based on random forest combining texture feature and NDVI

The accurate classification of sea ice is of great significance for mastering the growth and development of sea ice and ensuring the safety of navigation. Due to the influence of data sources and classification methods, the improvement of sea ice classification accuracy is limited. In this paper, for high spatial resolution optical remote sensing images, an accurate sea ice classification method based on texture features and normalized difference vegetation index (NDVI) was proposed, and a random forest classifier was used to construct a sea ice classification method. Taking Jiaozhou Bay of Qingdao as the experimental area and GF-2 as the experimental data, the sea ice types were extracted and compared with other classification methods. The results show that for GF-2 high-resolution optical remote sensing data, compared with the traditional random forest, support vector machine, automatic classification and regression tree methods and maximum likelihood classification method of combining texture features, the overall classification accuracy was improved by 13.70%, 11.60%, 19.22% and 29.37%, respectively. The Kappa coefficient was increased by 0.16, 0.13, 0.22 and 0.44, respectively. Compared with the random forest method based on texture features and normalized difference water index, the overall classification accuracy was improved by 9.67% and Kappa coefficient was increased by 0.09. It shows that the sea ice classification method constructed in this paper can effectively improve the accuracy of sea ice classification, and provide an effective technical means for the accurate classification of sea ice.