基于图论的图像分割算法仿真研究  

Image Segmentation Based on Graph Theory Algorithm Simulation Research

作　　者：张建梅[1] 孙志田[1] 余秀萍[1] 

机构地区：[1]河北建筑工程学院,河北张家口075024

出　　处：《计算机仿真》2011年第12期268-271,共4页Computer Simulation

摘　　要：研究图像分割优化问题。由于图像可以分割为若干个不同的区域,要求分割边缘清晰,速度快。但传统图像分割算法由于计算复杂等原因,造成图像分割分辨率低,清晰度不高,当图像中的信息量非常大时,分割非常耗时等缺陷,提出了图论的图像分割算法。采用图论的图像分割算法是一种全局的分割算法,首先分析图像在不同FRFT域的能量分布特点,通过归一化剩余误差因子ρ评估和分析FRFT域的能量积聚性和图像所包含的信息,使用最小生成树方法对图像对区域分别进行分割,并最终合并,采用二值化方法对图像进行仿真。结果表明,改进的算法能有效的分割图像,提高了图像分割的速度,是一种有效的新颖的图像分割算法。Images can be segmented into several different regions,and the segmentation edges must be clear enough and the speed must be fast.But the complex calculation and other reasons in the traditional image segmentation algorithms result in low resolution image segmentations,and the image clarities are not high.The put forward an algorithm for image segmentation.The graph theory image segmentation algorithm is a global segmentation algorithm.First the image energy distribution characteristics were analyzed in different FRFT domain.Then the method of normalization of residual error factor was uset to assess and analyze FRFT domain energy accumulation and the information of the image.The minimum spanning tree method was used for segment the image,and then eventually to merge the segments.The results show that,the algorithm can effectively improve the image segmentation and the segmenting speed,and is an effective and novel image segmentation algorithm.

关 键 词：图像分割 图论 最小生成树 仿真 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]