计算机工程 ›› 2011, Vol. 37 ›› Issue (21): 126-130.doi: 10.3969/j.issn.1000-3428.2011.21.043

• 人工智能及识别技术 • 上一篇    下一篇

高分辨率影像中基于纹理的建筑区信息提取

陈超祥,陈华锋,叶时平   

  1. (浙江树人大学信息科技学院,杭州 310015)
  • 收稿日期:2011-05-24 出版日期:2011-11-05 发布日期:2011-11-05
  • 作者简介:陈超祥(1975-),女,副教授、硕士,主研方向:智能计算,地理信息系统;陈华锋,讲师、博士;叶时平,教授
  • 基金项目:
    浙江省教育厅科研计划基金资助项目(Y200805711)

Building Area Information Extraction Based on Texture in High-resolution Image

CHEN Chao-xiang, CHEN Hua-feng, YE Shi-ping   

  1. (College of Information Science & Technology, Zhejiang Shuren University, Hangzhou 310015, China)
  • Received:2011-05-24 Online:2011-11-05 Published:2011-11-05

摘要: 在分析QuickBird高分辨率遥感影像特点的基础上,将纹理特征作为分类依据,通过对比实验得出参与分类计算的纹理特征参数为反差和增强反差,按标准距离选出最佳分类因子。选取绿色波段数据,使用改进的最小距离法自动提取影像中的建筑区信息。仿真结果表明,基于该方法提取的建筑区信息识别率为95.4%。

关键词: 遥感, 高分辨率, 纹理, 灰度共生矩阵, 建筑区信息提取

Abstract: Texture is introduced as basis of classification based on the analysis of QuickBird high-resolution remote sensing images. Contrast and enhanced contrast are determined suitable for classification through contrast experiments, and the best classification factors are determined according to standard distance. The green band is chosen as classification band, and construction area information is extracted by improved minimum distance. Simulation results show that the extraction accuracy reaches 95.4%.

Key words: remote sensing, high-resolution, texture, Gray Level Concurrence Matrix(GLCM), building area information extraction

中图分类号: