摘要
针对大型物体和回转体的三维形貌视觉测量,提出了一种基于公共特征点的三维数据拼接方法。该方法首先把测量对象的表面划分成若干个子区域,并且确保相邻的子区域含有重合区域,然后采用一套双目立体视觉测量系统对各个子区域进行单独测量,利用重合区域的公共特征点计算前后两次测量的空间变换矩阵,将三维点坐标转换到一个坐标系下完成拼接。变换矩阵的计算引入了Rodrigues参数,简化了运算过程,且具有较高的精度。该方法操作简单,非常适合于实际应用,实验结果证明了此方法的有效性。
The paper presents a 3D data registration method based on common feature points for large surface vision measurement. Firstly, the surface of the measured object is divided into several sub-regions and to ensure that adjacent sub-regions con- tain overlapping regions. And then 3D data of the measured object can be obtained in sub-regions by a binocular stereo vision measurement system. A certain number of common feature points in the overlap of sub-regions can be used to calculate the tran- sition matrix of the two measure space, So the 3D datas will be transformed into an unified coordinate frame. As a result, 3D reg- istration is completed. The Rodrigues parameters are introduced to calculate the transition matrix and make it simple and precise. The method is easy to be operated and very suitable for implementation. Experimental results show the effectiveness of this method.
出处
《微型机与应用》
2010年第6期39-42,共4页
Microcomputer & Its Applications
基金
广西科学基金项目(桂科基0991022
桂科基0731020
桂科能063006-5G-4)
