基于面向对象的养殖水面遥感分类识别

Extraction of Aquaculture Area Based on Objected-oriented Method

本文采用SPOT-5数据源,基于面向对象方法对养殖水面的识别与提取进行了研究。通过多尺度分割方法获取试验区的地物对象,针对不同地物的光谱、纹理、形状等信息,反复实验设置提取流程和规则,从而实现养殖水面的提取。与传统分类方法的对比,分类精度方面有了明显的提高,噪声也得到了明显的控制。

This research is focused on the identification and extraction of aquaculture area based on SPOT-5 image by using objectoriented classification method. Using multi-scale segmentation, and based on the spectral information, shape, texture etc, the process and rules are constructed. Comparing to conventional methods, it takes full advantages of feathers, which improves the accuracy of classification.