3D形貌测量中一种基于主成分分析的单帧投影栅算法

将主成分分析(PCA)引入投影栅形貌测量中,提出 了一种基于PCA的单帧投影栅算法。由计算机生成正弦条纹图,利用DLP投影仪投影到 被测物体表面,由CCD相机采集受物体 形貌调制的畸变条纹图并保存在计算机中进行处理。首先由采集的变形条纹图构成5帧 空域相移图；然后将多帧相移条纹图分解成不相关的主成分,由第1和第2主成分利用反正 切函数计算得到变形条纹相位信息；最后根据条纹相位的单调性,采用聚类分析法确定相位 全局符号。实验结果表明,本文算法无需确定相移量就可直接解调出与物体形貌对应的相位 数据,运算速度快,测量精度高,适用于在线三维形貌测量。

One-frame projection algorithm based on principal component analysis in 3D shape measurement

A one-frame projection algorithm is proposed to demodulate the phase by the principal component analysis (PCA).Sinusoidal fringe patterns generated by computer are projected onto the object surface by a digital-light-process ( DLP) projector and the deformed fringe patterns are acquired by a CCD camera.Firstly, five frames of phase-shifted fringe patterns are constructed from the ori ginal one-frame spatial carrier pattern.Secondly,the multiple frames of fringe patterns are decomposed into a set of values of uncorrelated variables called as pricipal components,thus the first and second principal components are obtain ed. Then,the phase is calculated from the actangent function of the two principal components.Finally,the correct global phase sign is determined by clustering method considering fringe phase monotonicity.The experimental results show that the proposed method can directly extract the phase distribution phase maps of the object without determining phase shifts.The algorithm is fast and accurate,which is suitable for on-line three-dimensional (3D) shape measuremen t.