基于GVF模型的多目标边缘提取  

Multi-object Edge Extraction Based on GVF Snake

作　　者：谢珊珊[1] 马社祥[1] 

机构地区：[1]天津理工大学计算机与通信工程学院,天津300191

出　　处：《光电工程》2010年第12期35-40,45,共7页Opto-Electronic Engineering

基　　金：国家自然科学基金资助项目(60872064)

摘　　要：针对活动轮廓模型的多目标边缘提取问题,本文提出了一种基于GVF模型的改进算法。该算法首先利用GVF模型对整个目标区域进行边缘提取,由于目标间存在滞止点区域,边缘提取的结果中将包含伪边缘。针对这些伪边缘进行分析并设置新的初始轮廓线,再次利用GVF模型进行目标边缘提取,并进行判断。将分别提取到的真实边缘首尾相连,去除伪边缘,可进行多次判断是否存在伪边缘,直至得到完整的多目标边缘。实验结果表明,该改进算法可以实现多目标边缘提取,且易于实现。An improved algorithm is presented based on the GVF Snake in order to achieve the multi-objective edge extraction. Firstly, the GVF Snake is used for carrying out the edge extraction of the whole object region, and the results will contain false edges because of the stagnation point region. Secondly, this false edge is analyzed and a new initial contour is established automatically. Then, the GVF Snake is used again to carry out new edges extraction and to determine the real edge and false edge. In the end, those real edges are connected, which were extracted in above two steps, and remove the false edges to get the complete multi-objective edge. Experimental results show that the proposed algorithm can achieve the multi-objective edge extraction.

关 键 词：主动轮廓模型 多目标 边缘提取 力场分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]