摘要
相干点漂移（Coherent Point Drift,CPD）方法是近几年出现的一类有效的点模式匹配方法,能对刚体和非刚体进行有效的匹配,但也存在计算效率较低及两个点集存在大角度旋转关系时匹配失效的问题。本文针对轮廓匹配问题,提出了一种融合点的位置和曲率特征的CPD模型的快速匹配方法。在匹配过程中,首先对目标轮廓和数据轮廓进行粗匹配,然后将粗匹配得到的变换参数作为CPD算法的初始输入参数,对轮廓上的仿射不变曲率尺度空间角点及角点在轮廓上的邻近点进行匹配,得到最终的变换参数。通过对实际图像提取的轮廓进行匹配实验,结果显示所提出的匹配方法具有较高的匹配精度,能实现图像的精确配准并能显著提高匹配速度。
Coherent Point Drift is a kind of effective Point pattern matching method appeared in recent years,which can effective match rigid body and non-rigid body. However,this method also has disadvantages in low computational efficiency and matching failure when the two point sets have a large rotation relationship. Aiming at the problem of contour matching,a fast matching method of CPD model fusing the position and curvature characteristics is proposed.In the matching process,the target contour and data contour are first rough matched,and the matching parameters are obtained as the initial input parameters of the CPD algorithm,the final transformation parameters are obtained by matching the affine invariant curvature scale space corners and their adjacent point in the contour. The matching experiments on the actual image show that the proposed method has a high matching precision and significantly improves the matching speed.
作者
贺飞跃
贺兴时
赵文芝
HE Feiyue;HE Xingshi;ZHAO Wenzhi(Xi＇an Engineering University School of Science,Xi ＇an,710048)
出处
《自动化与仪器仪表》
2018年第8期145-148,共4页
Automation & Instrumentation
基金
国家自然科学基金青年科学基金项目(No.11501436)
陕西省教育厅专项科研计划项目(No.16JK1326,No.17JK0340)
西安工程大学博士科研启动基金项目(No.BS1420)
关键词
轮廓
匹配
仿射不变角点
相干点漂移
contour
matching
affine-resilient corner
coherent point drift
