海岸线变化是研究海岸带演化及全球气候变化的基础之一, 而从遥感影像上快速、准确地自动提取不同类型 的海岸线一直是个难点, 尤其是淤泥质海岸的岸线。采用面向对象的方法, 针对苏北淤泥质海岸地区2006 年的 T M 影像, 建立包括光谱变换信息和纹理特征信息的规则集对影像进行高精度分类, 进而准确地提取海岸线。使用 该方法不但具有充分挖掘影像信息、自动化程度高、提取速度快等优点, 而且可以从精度和稳定性两个指标对提取 结果进行检验, 不依赖于参考岸线, 检验结果较为客观。

The coastline chang e is one of the research bases of co ast al evo lutio n and g lobal climate change. Automatic ex tractio n of different coastline ty pes from the remote sensing images quickly and accur ately is alwa ys difficult, especially for the sho reline of muddy co ast. In t his paper , the o bject2or iented method was adopted to build the rule sets including the spectrum t ransfo rm in2 fo rmation and text ur e feat ur e informat ion, to classify the imag es in hig h definit ion, and then to ex tr act the coastline accurat ely based o n t he T M remot e sensing imag es of muddy coast zone in no rthern Jiangsu in 2006. The r esult s show ed that the method can ex cav ate the image info rmatio n suff icient ly w ith a high deg ree o f automat ion and fast ex tr action, and also test the r esults from t wo index es o f accuracy and stabilit y independently of the r efer ence coastline.

国家公益性行业科研专项( 201011019)