论文部分内容阅读
以山西省古交市嘉乐泉乡为试验区,采用SPOT-5的10m、5m和2.5m 3种影像数据对退耕还林地面积进行分类监测。所设计的2种方案分别是:1)将地物类型分为7类,退耕还林地作为一种单独地类,对3种影像数据进行计算机自动分类和2.5m影像的人工解译分类;2)借助退耕还林作业设计图,将退耕还林地块影像分割出来,对退耕还林地和未退耕还林地进行有监分类。精度验证表明,第一种方案中2.5m融合图像的人工解译分类,退耕还林地的分类精度在50%以下;第二种方案中3种影像数据的总体分类精度均大于90%。建议在退耕还林地的作业设计图电子化的基础上,应用SPOT-5数据监测退耕还林地的任务完成和植被覆盖情况。
Taking Jialequan Township, Gujiao City, Shanxi Province as the experimental area, the land area of returning farmland to forestry was classified and monitored by using three kinds of image data of 10m, 5m and 2.5m of SPOT-5. The two schemes designed are as follows: 1) The terrain types are divided into 7 categories, and the conversion of cropland to forestland is a separate category, and the three kinds of image data are automatically classified by computer and manually interpreted by 2.5m images.2 ) With the design of returning farmland to forest project, the image of the converted farmland to forestland will be segmented, and there will be supervised classification of returning farmland to forestland and non-returning farmland to forestland. Accuracy verification shows that the classification accuracy of the 2.5 m fusion image in the first scenario is less than 50% and the classification accuracy of the three kinds of image data in the second scenario is greater than 90%. It is suggested that SPOT-5 data should be used to monitor the completion of the conversion of cropland to forestland and the vegetation cover based on the electronic design of the plan of returning farmland to forestland.