English  |  设为首页  |   加入收藏
 
 
面向对象多特征融合的水域岸线目标变化检测------张 曦,王春林,黄祚继,董丹丹
      来源:南京水利水文自动化研究所    [大][中][小]

  曦,王春林,黄祚继,董丹丹

(安徽省(水利部淮河水利委员会)水利科学研究院,安徽 合肥 230001

 : 为认真落实河长制“清四乱”等专项行动,量化水域岸线监管测评工作,以岸线码头为目标,研究一种基于面向对象思想多特征融合的水域岸线目标变化检测方法。针对多时相高分辨率遥感影像,利用面向对象多尺度分割原理将具有空间连续性的同类区域划分为目标对象,提取目标对象的光谱、纹理及几何结构组成特征矩阵,并利用高斯径向基核函数支持向量机(RBF-SVM)进行分类;计算变化矢量差值,并与人工判读数据对比分析得到目标变化检测结果。实验结果表明,该研究应用于水域岸线上目标的变化检测中效果明显,RBF-SVM分类误差影响最终目标变化检测的正确率,可为实现河湖水域岸线长效管护提供技术支撑。

关键词:面向对象;多特征融合;支持向量机;水域岸线管护;变化检测

Detection of shorelines water bodies change using object-oriented method based on multi-feature fusion

ZHANG XiWANG Chunlin, HUANG Zuoji, DONG Dandan

(Water Conservancy Research Institute of Huaihe River Water Conservancy Commission, the Ministry of Water Resources, Anhui Province, Hefei230001, China)

Abstract: To implement the specific project of river chief system, clean up the chaos and quantify the measurement and evaluation of shorelines of water bodies, this research presents an object-oriented method for shorelines of water bodies change detection based on multi-feature fusion. From multi-temporal high spatial resolution remote sensing imagery, this paper first divides the image into spatial continuous objects using a multi-resolution segmentation algorithm, and extractes the spectral, texture and geometric features of the objects to generate the feature matrix. The RBF-SVM method is used to classify the multi-temporal imagery. This paper calculates the vector difference between the two classified images and compares the change detection result by manual operation. The result shows that the research has an obvious effect for shorelines of water bodies change detection and the RBF-SVM classification method enhances the change detection accuracy. It provides technical support for the long-term management and protection of the shorelines of rivers and lakes.

Key words: object-oriented; multi-feature fusion; support vector machine; management and protection of shorelines of water bodies; change detection
面向对象多特征融合的水域岸线目标变化检测.pdf (3.91 MB)
2020-03-02 17:31

lixinling2020-03-02浏览(545   [关闭]
   

    

English  |    设为首页  |   加入收藏  

 

 

 
 
进入编辑状态