计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (22): 222-225.

• 工程与应用 • 上一篇    下一篇

融合多尺度分割与CART算法的矸石山提取

赵  慧1,2,汪云甲1,2   

  1. 1.中国矿业大学 环境与测绘学院,江苏 徐州 221116
    2.中国矿业大学 国土环境与灾害监测国家测绘局重点实验室,江苏 徐州 221116
  • 出版日期:2012-08-01 发布日期:2012-08-06

Fusing muti-scale segmentation into CART algorithm on coal gangue extraction

ZHAO Hui1,2, WANG Yunjia1,2   

  1. 1.School of Environment Science and Spatial Informatics, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
    2.Key Laboratory for Land Environment and Disaster Monitoring of SBSM, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
  • Online:2012-08-01 Published:2012-08-06

摘要: 结合多尺度分割和CART算法的特性,提出一种新的目标信息提取方法。其基本思想是将小尺度分割与大尺度分割相结合,将影像分割成一系列同质性对象;以同质性对象为基本单元选择训练样本,后利用CART算法提取目标信息。实验结果表明:与单纯像素级的CART算法相比,该方法可有效减少提取结果的噪声,一定程度上排除了其他地类对目标信息的干扰,提取精度显著提高。

关键词: 多尺度分割, 分类和回归树(CART), 矸石山, 目标提取

Abstract: Considering the properties of muti-scale segmentation and CART, a method for information extraction is proposed in this paper. Integrating small-scale segmentation with big-scale segmentation, the remote sensing image is segmented into a series of homogenous regions through watershed transformation and region merging process. The region-based information extraction is performed by utilizing CART after sample selection. Experimental results indicate that the proposed information extraction method depresses the image noise effectively, the extraction accuracy is improved and more interpretable result can also be achieved.

Key words: muti-scale segmentation, Classification and Regression Tree(CART), coal gangue, information extraction