三维图像拼接技术是实现大型物体形貌测量的关键。本文在图像控制点约束的基础上,提出了一种实现三维图像拼接的新方法,即将一伪随机空间编码投射到一大型物体表面,然后用CCD在不同视角拍摄不同视场中的物体,利用重叠区域相同的某一窗口特征点进行配准,然后采用四元组法求取坐标转换矩阵,从而实现大范围自由曲面三维形貌的测量。实验结果表明:应用空间编码的图像拼接新方法可以实现大尺度三维面形的测量,拼接数据相对误差<0.8%。

3D image merging is a key technique in large scale free-form surface measurement.Based on image-control-point-restraint theory,a new 3D image merging method is presented.In the method,a pseudo-random space code is firstly projected on a large object surface,and the images of the object are taken from different angles of view.Through matching the interest points on the overlapping region,the coordinate transformation matrix is obtained and 3D measurement of large-scale free-form surface can be realized.The experimental results show that this new image merging method by space code projection can measure large-scale 3D shape and the relative error of merging data is less than 0.8%.