# 基于高分一号卫星遥感数据提取城市建设用地方法研究Research on the Method of Extracting Urban Construction Land Based on Gaofen-1 Satellite Remote Sensing Data

• 全文下载: PDF(1546KB)    PP.334-340   DOI: 10.12677/AG.2019.95037
• 下载量: 65  浏览量: 124   科研立项经费支持

The phenomenon of urban sprawl and contraction in cities seriously restricts the coordinated regional development. It is urgent to extract the boundary of urban buildings by remote sensing method to assist the government in policy decision and coordinate regional development. This paper selects supervision classification, support vector machine classification and vegetation in-dex classification, extracts urban construction land based on gaofen-1 satellite remote sensing image, and discusses the construction land extraction method suitable for gaofen-1 remote sens-ing image through precision test. The results show that support vector machine classification has the best effect on gaofen-1 image extraction, the average effect on supervised classification and the poor effect on vegetation index classification.

1. 引言

2. 研究区域

3. 分类方法

3.1. 植被指数法

$\text{NDVI}=\left(\text{}IR-R\right)/\left(IR+R\right)$ (1)

3.2. 监督分类法

$P\left(\frac{{n}_{i}}{\omega }\right)=\frac{P\left({n}_{i}\right)P\left(\frac{\omega }{{n}_{i}}\right)}{P\left(\omega \right)}$ (2)

3.3. 支持向量机法

$\begin{array}{l}H:w*x+b=0\\ {H}_{1}:w*x-b=1\\ {H}_{2}:w*x-b=-1\end{array}$ (3)

3.4. 建设用地判定

Figure 1. Gaofen no.1 (WFV) image construction land (left: true color, right: false color)

4. 分类结果与分析

4.1. 图像预处理

Figure 2. Original image of the study area

4.2. 植被指数分类

Figure 3. Diagram of vegetation index classification results

4.3. 监督分类

Figure 4. Diagram of supervised classification results

4.4. 支持向量机分类

Figure 5. Diagram of support vector machine classification results

4.5. 分类比较

Table 1. Comparison of land extraction accuracy of Gaofen-1 satellite

5. 结论

1) 运用支持向量机方法和监督分类方法提取高分一号影像建设用地效果较佳，kappa系数和总体精度分别为0.8295、91.42%与0.7513、87.39%，均达到分类精度要求，错分较少，图斑规则，联系度较高，但监督分类结果仍具有较多的细小斑块；运用指数分类方法提取建设用地效果差，kappa系数和总体精度分别为0.6769、80.24%，未达到分类精度要求。

2) 对于高分一号遥感影像，一般采用支持向量机法对区域范围较小，建设用地地类斑块较集中的地区进行建设用地的提取，采用监督分类方法对区域范围较大，建设用地几何、光谱特征较明显的地区进行建设用地提取；一般不采用植被指数法进行建设用地的提取。

 [1] 张华, 练云龙. 国外收缩城市研究进展及其启示[J]. 西部人居环境学刊, 2018, 33(3): 34-42. [2] Bhaskaran, S., Paramananda, S. and Ramnarayan, M. (2010) Per-Pixel and Object-Oriented Classification Methods for Mapping Urban Features Using Ikonos Satellite Data. Applied Geography, 30, 650-665. https://doi.org/10.1016/j.apgeog.2010.01.009 [3] Huertas, A. (1988) Detecting Buildings in Aerial Images. Computer Vision Graphics & Image Processing, 41, 131-152. https://doi.org/10.1016/0734-189X(88)90016-3 [4] 齐永菊, 裴亮, 叶国凤, 等. 高分辨率遥感影像建筑物提取方法研究[J]. 测绘与空间地理信息, 2018(2): 119-123. [5] 陈杰. 高分辨率遥感影像面向对象分类方法研究[D]: [博士学位论文]. 长沙: 中南大学, 2010. [6] 东方星. 我国高分卫星与应用简析[J]. 卫星应用, 2015(3): 44-48. [7] 马荣华, 陈雯, 陈小卉, 等. 常熟市城镇用地扩展分析[J]. 地理学报, 2004, 59(3): 418-426. [8] 刘经南, 姚宜斌, 施闯. 基于等价方差–协方差阵的稳健最小二乘估计理论研究[J]. 测绘科学, 2000, 25(3): 1-6.