论文部分内容阅读
以云南大学呈贡校区作为研究区,使用高分辨率的GeoEye-1数据,利用智能化影像分析软件eCognition,采用多尺度分割方法和最邻近法,探索性的进行面向对象的土地覆被分类研究。研究成果显示使用这种面向对象的土地分类方法得到的土地覆被类型判识精确度高达90%,颜色信息丰富、形状接近真实地物、大小区分明显、纹理信息突出、上下文关系明确、总体分类精度达到90%以上。同时研究成果还为呈贡校区的土地绿化建设和重点部署提供了技术支撑。
Taking Chenggong campus of Yunnan University as a research area, this paper uses GeoEye-1 data of high resolution and eCognition of intelligent image analysis software to explore object-oriented classification of land cover using multi-scale segmentation and nearest neighbor method . The research results show that using this object-oriented land classification method, the recognition accuracy of land cover types is as high as 90%, the color information is rich, the shape is close to the real features, the size is obvious, the texture information is prominent, the context is clear, the overall classification Accuracy of more than 90%. At the same time, the research results also provide technical support for the land afforestation construction and key deployment of Chenggong campus.