Study On The 3-D Shape Measurement Of Physical Simulation Experiment Model

The application of three dimensional (3-D) shape measurements isimportant in the area of machine vision, on-line product inspection, geologicprospecting, etc. Fourier transform profilometry is widely used among all themethods of 3-D sensing.This paper is one of the important parts of geologic structure modelingexperiment, and is aimed to get the accurate outline of the anamorphic object.The key techniques like filtering, phase unwrapping and system calibrationwere investigated in this paper.In Fourier transform profilometry, the frequency will be dispersed andconfused when detecting complex objects. It is difficult to filter out theinutility information. In this paper, Butterworth, Hanning and Rect filter aswell as the wavelet methods were chosen to analyze the accuracy of FTPwhen noise exists and does not exist. The error distribution gave a conclusionthat the Hanning filter has a better restrain on noise, and the wavelet analysisimproved a lot both in measuring precision and range as it can analyze thesignal both in time and frequency domain.In 3-D shape measurement, the wrap phase will not be unwrappedcorrectly when it was polluted by noise or shadows of the object. Two kinds of phase unwrapping algorithms were studied: path-following andminimum-norm methods. In this paper, we used path-following method tounwrap actual objects measured by project, and got the conclusion that thephase derivative variance is generally the most reliable measure of phasequality, while the maximum phase gradient correctly marked thediscontinuity domain.System calibration is essential in optics 3-D shape measurement. Z and(X,Y) coordinate calibration are two parts of it, in which Z calibration meansto find out the relationship between height and phase difference, while (X,Y)calibration is to build a correspondence between image coordinate and worldcoordinate system. We studied the mathematical model of camera imaging,introduced a calibration method for 3-D machine vision, put forward acalibrate error analysis, and the result showed a higher acquisition andefficiency of this method.Theoretical analysis and practical measurement prove the availability of allof the methods and algorithms.