基于GF-1卫星影像的临洪河口湿地遥感分类  

Classification of Linhong River Estuary Wetland Based on GF-1 Satellite Image

作　　者：费鲜芸 何鑫坤 谢宏璇 王灏旻 陈实 陈天羽 FEI Xianyun;HE Xinkun;XIE Hongxuan;WANG Haomin;CHEN Shi;CHEN Tianyu(School of Marine Technology and Geomatics,Jiangsu Ocean University,Lianyungang 222005,China)

机构地区：[1]江苏海洋大学海洋技术与测绘学院,江苏连云港222005

出　　处：《江苏海洋大学学报（自然科学版）》2021年第2期50-57,共8页Journal of Jiangsu Ocean University:Natural Science Edition

基　　金：国家自然科学基金资助项目(31270745)。

摘　　要：滨海湿地的遥感分类研究对于滨海湿地的保护和规划具有重要意义。GF-1号卫星是我国高分卫星系列中可获取性较好的卫星,对其开展影像数据的湿地分类方法研究有助于获得更为详尽的滨海湿地类型等信息。以连云港临洪河口为例,利用GF-1号遥感影像进行预处理后,经过影像分割,基于面向对象原理在确定最优分割尺度参数上采取随机森林模型,控制灰度共生矩阵提取的纹理特征为实验变量,对滨海湿地的土地利用类型分类,并比较纹理特征在随机森林模型中对分类精度的影响。结果表明,添加纹理特征后的总体精度达到89.75%,Kappa系数达到0.88,相比未添加纹理特征的分类精度提高了7%左右。因此,GF-1号影像适用于滨海湿地区域土地利用类型的分类,纹理特征可以有效提高其分类精度。The remote sensing classification research of coastal wetlands is of great significance to the protection and planning of coastal wetlands.The GF-1 satellite is the most accessible satellite in my country’s high-resolution satellite series.Research on wetland classification methods for its image data will help to obtain more detailed information about coastal wetland types.This paper takes the Linhong River Estuary in Lianyungang as an example.After preprocessing the GF-1 remote sensing image,the image is segmented,and based on the object-oriented principle,a random forest model is used to determine the optimal segmentation scale parameters to control the gray level co-occurrence matrix(GLCM)extraction.Texture features are experimental variables to classify the land use types of coastal wetlands,and compare the effects of texture features on the classification accuracy in the random forest model.The results show that the overall accuracy after adding texture features reaches 89.75%,and the Kappa coefficient reaches 0.88,which is about 7%higher than the classification accuracy without adding texture features.It can be seen that the GF-1 image is suitable for the classification of land use types in coastal wetland areas,and the texture feature can effectively improve its classification accuracy.

关 键 词：高分1号卫星 纹理特征 灰度共生矩阵 滨海湿地 随机森林 

分 类 号：P23[天文地球—摄影测量与遥感] TP753[天文地球—测绘科学与技术]