基于面向对象的青海湖环湖区居民地信息自动化提取

居民地的空间格局和密度直接反映着区域人类活动的强弱程度，影响着区域人地系统演变和生态环境可持续发展。基于高分辨率卫星遥感影像数据，提出了一种面向对象的青海湖环湖区居民地信息自动化提取方法。首先，利用尺度集理论对高分辨率卫星遥感影像进行多尺度分割，获取不同尺度的分割对象；其次，通过机器学习算法集对分割对象的自定义特征、光谱特征、几何特征和纹理特征进行训练，选取最优自动分类算法；最后，利用最优自动分类算法提取青海湖环湖区城镇居民地和农村居民地信息。采用平均召回率、平均准确率和平均F值评价指标对分类结果进行精度评价，其中，城镇居民地各评价指标均在93%以上，农村居民地各评价指标均在86%以上。结果表明:该方法提取城镇居民地和农村居民地总体精度较高，在大面积人类活动精细化监测中具有较好的科学意义和应用价值。

Automatic Extraction of Residential Information based on Object-oriented in the Areas around the Qinghai Lake

The spatial pattern and density of residential areas directly reflect the intensity of regional human activities, and affect the evolution of a regional human-land system and the sustainable development of ecological environment. In this study, we proposed an objected-oriented automatic extraction method, which based on the high spatial resolution satellite remote sensing image data in the surrounding area of Qinghai lake watershed. Firstly, multi-scale segmentation of high-resolution satellite remote sensing image was carried out by using the scale sets theory to obtain segmentation objects in different scales. Secondly, the custom, spectral, geometric and texture features of the sample attributes were trained through the sets of machine learning algorithms, and the optimal automatic classification algorithm was selected. Finally, the optimal automatic classification algorithm was used to extract the information of urban and rural residential areas in the surrounding area of Qinghai lake watershed. The average recall rate, accuracy rate and F value were used to evaluate the classification results. Accuracy evaluation indexes of urban residential areas were more than 93%, and those of rural residential areas were more than 86%. The results show that this method has higher overall precision when extracting urban residential areas and rural residential areas, and has better scientific significance and application value in fine monitoring of human activities in large areas.