数学形态学在空间格局图像骨架提取中的应用  

Application of Mathematical Morphology in the Skeleton Extraction of Spatial Pattern Image

作　　者：杨志平[1] 齐清文[2] 黄仁涛[1] 

机构地区：[1]武汉大学资源与环境科学学院,武汉430079 [2]中国科学院地理科学与资源研究所,北京100101

出　　处：《地球信息科学》2003年第2期79-83,87,共6页Geo-information Science

基　　金：863信息领域对地观测数据处理与分析技术项目 (2001AA135140);中国科学院知识创新工程领域前沿项目 ( CXIOG-D001-01);863信息领域空间信息应用与产业促进专题项目 (2002AA134076)

摘　　要：本文从遥感图像上提取空间格局的过程中 ,对基于数学形态学的主要细化算法进行了归纳总结 ,结合各算法实现的效果图 ,将其进行对比分析 ,找出它们各自的优缺点。最后对其进行了综合改进 ,针对空间格局图像的复杂多变、非细长的面状特征 ,提出一种提炼其骨架的细化算法 ,并且给出了具体的程序实现方法。Mathematical morphology is a powerful implement in the research of image analysis.It has been a primary domain of image science and its theory and method has been applied in many fields.The article is to study its method in the process of extracting the skeleton of spatial pattern based on remote sensing data.Spatial pattern image differs from area with slab-sided shape such as character and fingerprint.It is not slab-sided with daedal orientation.The authors sum up the current major mathematical morphology arithmetic of thinning at first.They find out the merits and demerits by analysing the difference between these arithmetic using the resultant contrastive photos.Finally, based on these arithmetic, the authors bring forward an improved arithmetic of skeleton extraction of spatial pattern image that is very fit for not slender bar with large area, furthermore offer the realization method for the program.

关 键 词：数学形态学 结构算子 击中击不中 最大数字圆盘 

分 类 号：P23[天文地球—摄影测量与遥感]