Study On Calibration Of Profile Measurement Though Array Stripes Produced By Laser Interference

In modern production and life,computer technology has developed rapidly.It has been widely used in all fields,which accelerates development of many other technologies.Machine vision technology is a typical one.Machine vision technology grants the machines human visual function and makes the machines work like human eyes.Compared with traditional measuring methods,it's non-contact,cheap,precise and welladapted.Due to its high research significance and great prospect,it has been used in industrial manufacture,military,medical treatment,archaeology and so on.The major methods to do machine vision are divided into two different types.One is surface structured light projection,and the other is linear structured light scanning method.These two methods have advantages over traditional measuring methods,but they all have their weaknesses.Surface structured light projection is expensive because of the optical grating and projector which should be precise.Meanwhile,getting thin fringes is hard.It's very difficult to do some precise measurement.A machine to move the object precisely is needed for linear structured light.It also costs some time to move the object,so this method can't be used in real-time measurement.In conclusion,the existing methods all have their defects.In this thesis,a new way to achieve machine vision is proposed.It can solve the problems above.The project in this thesis is designed as follows.A laser beam reflects twice on the front surface and the rear surface.Then interference fringes are produced on the surface of the object.An industrial camera collects the image and delivers it to the computer.The program processes Fourier band-pass filter and thinning,so the information in these interference fringes is conversed into data and saved.Then image calibration at several different positions is accomplished.These calibration parameters are highly related,and they are determined by marked positions.A mathematical formula to match the relationship between calibration parameters and these marked positions is needed.After getting this matching formula,we can get calibrating parameters at all positions.With the help of matching formula and the data of these fringes,we can get three-dimensional coordinates of all the coordinates in the image captured by a camera.Finally the 3-D profile of the object is shown using OpenGL.In this thesis,image processing should be carried out,including laser lock-in image,Fourier bandpass filter,thinning and stripe segmentation.Laser lock-in image separates background lights from the image.Then Fourier bandpass filter clears up noise frequency.Next,thinning of a stripe is achieved by using an extremum algorithm.Finally stripe segmentation stores the data of every stripe.