基于Hough变换的椭圆检测算法对比分析
2018-12-10成浩崔文超
成浩 崔文超
摘要:椭圆检测在图像识别与计算机视觉領域具有重要研究意义。为了对比分析出不同椭圆检测算法的优缺点,提高检测精确度和效率,针对基于Hough变换的椭圆检测4种算法(传统Hough变换方法、几何特征椭圆检测法、弦中点椭圆检测法和三点椭圆检测法)进行仿真,并对不同算法进行有效性分析。通过对检测结果进行客观评价,得出对比结论,并根据实验结果分析不同算法适用的对象。设计了椭圆检测对比分析软件平台,便于进行不同算法的对比分析。
关键词:椭圆检测;Hough变换;对比分析
DOIDOI:10.11907/rjdk.182095
中图分类号:TP312
文献标识码:A文章编号文章编号:16727800(2018)009011504
英文标题Contrastive Study of Elliptical Detection Based on Hough Transform
——副标题
英文作者CHENG Hao, CUI Wenchao
英文作者单位(College of Computer and Information Technology, China Three Gorges University, Yichang 443000,China)
英文摘要Abstract:Ellipse detection is of great importance in image recognition and computer vision. In this paper, in order to compare and analyze the advantages and disadvantages of different ellipse detection methods and improve the accuracy and efficiency of detecting ellipse from images,four Hough transform based ellipse detection methods have been simulated, including traditional Hough transform method, geometric feature ellipse detection method, string midpoint ellipse detection method and threepoint ellipse detection method. The detection effectiveness regarding different algorithms has been analyzed and tested through the objective evaluation on the experimental results. Finally, the relevant contrastive conclusions are drawn, and the applicable objects of different methods are analyzed. At the same time, a software platform for ellipse detection and comparative analysis is designed to facilitate the comparative analysis of different algorithms.
英文关键词Key Words:ellipse detection; Hough transform; contrastive analysis
0引言
随着计算机视觉技术发展,人们可用计算机从图像中确定椭圆的位置和形状大小,如在数字摄影测量领域, 椭圆检测可提高图像匹配自动化程度和匹配率[1];在模式识别领域,人脸识别和行人头部的轮廓提取均可采用椭圆检测技术实现[2]。因此,椭圆检测有良好的应用前景,设计椭圆检测对比分析系统对椭圆检测算法的对比分析具有重要意义。
椭圆检测算法主要分为投票(类聚)和最优化两大类。最优化例如最小二乘法[35],其通过最小误差的平方和寻找符合样本点的最佳函数,此类算法准确度高,但需要预先对数据进行分组处理,不能同时检测多个椭圆。基于Hough变换的算法[69],运用两个坐标空间之间的变换,将样本点映射到参数空间,然后采用统计投票或建立累加器的方法,在参数空间中寻找峰值确定椭圆。……
