摘要
先通过几何矫正、配准、裁切等预处理获取深圳市辖区的HJ-1-A/B卫星CCD2数据,再选取城市绿地、林地、农用地等样本进行光谱特征分析,并以深圳市基本生态控制线等数据作为辅助,最后利用"决策树-监督"混合分类模式提取深圳城市绿地信息。将分类结果与深圳市生态资源测算的城市绿地面积进行对比,分析发现两类数据的Pearson相关系数值为0.9997,平均误差为8.37%;通过空间分布对比发现,两类数据解译获取的结果基本一致。结果有利于提高城市绿地信息的提取效率,进而为降低区域生态环境动态监测的成本并提高监测的时间精度提供一定的借鉴。
Through pretreatments including geometric correction, registration and cutting, A, B satellite data of Shenzhen, spectral characteristics were analyzed by samples of green space, forestland and farmland plots. Based on data of basic ecological line, green space information of Shenzhen was extracted through a hybrid classification model of decision tree and supervised method. The Pearson Correlations is 0.9997 between the classification results and those of ecological resource calculation program of Shenzhen, and the average error is 8.37%. The spatial distribution results were almost the same on spatial distribution. These results of the study will be helpful to improve the efficiency of extracting green space information, and could provide a certain reference for reducing the cost and raising time precision of regional eco-environmental dynamic monitoring.
出处
《生态科学》
CSCD
北大核心
2014年第3期568-573,共6页
Ecological Science
基金
2012年度深圳市生态资源状况测算项目(0722-1261-FE445SZB)
2013年度生态资源测算项目(SZZZ2013C0011)
大鹏新区生态文明示范区建设规划(SZCG2013040569)
关键词
CCD2影像
城市绿地
决策树分类
监督分类
CCD2 image
green space
decision-tree classification
supervised classification
