基于高分辨卫星影像的林业信息提取方法研究

Research for forestry information extracting method based on high resolution satellite images

在“生态高地”成为最美底色,保护森林资源的背景下,针对林地信息提取问题,采用国产高分辨率卫星影像高分二号(GF-2)数据,利用改进指数特征的面向对象分类算法进行林业信息的高精度分类。研究结果表明:该方法能有效提取向阳面林地、低覆盖度林地等复杂林业信息,以及完整的人工设施。通过使用预留验证样本进行精度分析,证实了改进后的面向对象分类方法相较于传统方法具备显著的精度优势,为林地管理和生态保护提供了有力的数据支持。

Against the backdrop of the most beautiful ecological highland and the protection of forest resources,this paper focuses on the problem of forest information extraction by using domestically produced high-resolution satellite image Gaofen No.2 data and an improved index feature based object-oriented classification algorithm to perform the high-precision classification of forestry information.The research results indicate that using this kind of method can effectively extract complex forestry information such as sunny forest land and low coverage forest land,as well as complete artificial facilities.By using reserved validation samples for accuracy analysis,it was confirmed that compared with traditional methods,the improved object-oriented classification method has significant accuracy advantages which provide strong data support for forest management and ecological protection.