基于Mexican Hat小波变换和区域生长的光学轮廓术的研究

Study on optical profilometry based on Mexican Hat wavelet transformation and region growing

提出了一种新的基于结构光投影的光学轮廓术的测量算法。该算法利用Hilbert变换和Mexican Hat小波变换进行相位提取,利用小波变换的平滑功能实现初步滤噪和消除高次谐波。然后利用基于截断相位差统计分布和区域生长算法进行解包,并通过对无效点的判断和解包后的邻域平滑实现了准确的相位解包。文中通过仿真计算证明了该方法较其他光学轮廓术有更强的噪声抑制能力,并实验验证了其较高的测量精度。

A novel measurement algorithm in optical profilometry based on structure light projection is proposed. The wrapped phase is retrieved by Hilbert transformation and Mexican Hat wavelet transformation, whose smoothing function can eliminate the elementary noises and higher order harmonics. The unwrapped phase is precisely achieved by the judgement of invalid points and filtering based on statistical distributions of wrapped phase difference and region growing algorithm. This method is demonstrated to have much better performances in suppressing noises and obtaining high measuring accuracy compared with other approaches by simulation and experiments.