森林资源年度监测小班数据自动更新技术

Automatic updating technique of subcompartment data for annual monitoring of forest resource

以相邻两年度遥感影像和上一年度小班数据为基础,研究了小班数据自动更新技术。通过多时相遥感特征与植被变化相关性分析,筛选出对判别植被变化贡献较大的一些遥感特征,提取每个小班的遥感特征,对现势的遥感影像和小班历史GIS数据的先验知识进行综合分析,建立森林资源变化判别规则,以确定变化小班。然后以变化的小班内遥感影像为研究对象,分别利用边缘提取和图像分割方法自动提取变化界线,产生分割线,再用分割线更新小班界线,从而生成本年度森林资源空间数据。研究表明:基于边缘提取方法的自动更新结果存在较多的伪边界,同时又丢失了一些真正的边界,其自动更新效果不理想;而基于图像分割方法自动更新的小班变化界线与人工目视勾绘的小班变化界线基本一致,可以满足生产要求。

Based on adjacent bi-annual remote sensing imageries and the associated subcompartment data of last year, automated developing techniques for the subcompartment data was presented in this study. By correlation analysis of multi temporal remote sensing features and vegetation change,certain remote sensing features that might make a greater contribution to the identification of actual vegetation changes were picked up. Combining these selected remote sensing features with prior knowledge of historical subcompartment GIS data, the discrimination method for forest resources changes was established through synthetically analysis, and then the changed boundaries for each subcompartment were extracted by both edge extracting and image segmentation separately method. Subcompartment edges were finally clipped with the generated boundaries,and the forest resource data updating of this year was finished. The results indicated that improper certain pseudo boundaries exist and true boundaries lost when the edge extraction techniques emploved. However,change information extracted by image segmentation algorithms presented a satisfactory result in accordance with the production requirements.The result derived from image segmentation methodologies agree well with those obtained by artificial visual delineation.