面向对象的遥感影像湿地分类与变化检测研究

The Research of Classification and Change Detection of Wetland Based on Object-oriented Remote Sensing Image

在面向对象技术支持下,运用多特征融合的遥感影像分类方法进行湿地分类提取,并通过对比得出两个时期的湿地变化.一时期Aster影像的总体精度为87%,Kappa系数为0.85;二时期GF影像的总体精度为87%,Kappa系数为0.83.通过对比计算发现福州市龙祥岛地区湿地总量减少,变化较显著的两类湿地是河流和滩涂,河流面积增加5.54%,滩涂面积减少6.83%.为及时掌握湿地的情况、做好湿地的开发与保护提供了参考.

With support of object - oriented technology, the article uses remote sensing image of multi - feature fusion wet-land classification and extraction to get the result of wetland change in two periods. The overall accuracy of the Aster image in the first period is 87% and the Kappa coefficient is 0. 85 ; the overall accuracy of the GF image in the second period is 87% and the Kappa coefficient is 0. 83. By comparison and calculation, the research finds that the total wetland area is re-duced in Long Xiang Island of Fuzhou city, and there is a significant change in the rivers and beaches, with an increase of 5.6% in rivers area and 6. 8% reduction in beach area. This study can grasp the wetland situation, provide some decision support for the development and protection of wetland.