Research On Object 3D Reconstruction Based On Grating Projection

Raster projection technology has attracted much attentions in the field of 3D reconstruction and is highly applied in industrial production lines,medical diagnosis,intelligent driving and face recognition.It is now the most widely used active non-contact 3D measurement technologies thanks to its simple structure,accurate detection,fast speed and low requirement for working environment.This dissertation with the benefit of experimental verification platform,studies on the three-dimensional reconstruction on the basis of theoretical research and simulation experiment.Through the vision measurement system of principle of grating projection model,the mapping relationship between image phase and object height is established,which lays the foundation for the research of 3D reconstruction system.The simulation experiment,comparison and analysis of four phase value algorithm--three,four,five step phase shift algorithm and Fourier Transform Profilometry,shows that four step phase-shift method is simple to operate with less experiment error.To obtain the continuous phase,the paper researched several frequently-used Phase unwrapping methods,for example,the gray code,the multifrequency method,cutting method and the least square method.Also,it put forward a diamond phase unwrapping algorithm based on above traditional algorithms and DCT.Through principle analysis and simulation experiment,it is concluded that,compared with the traditional algorithm,its reconstruction effect is better with the phase unwrapping diagram improved significantly.The dissertation studied the camera imaging process and the nonlinear model,adopted the Zhang checkerboard method for camera and projector calibration work,and applied MATLAB and OpenCV to carry out camera calibration comparison experiment,which shows that the calibration based on OpenCV calibration method has higher accuracy.Experimental verification platform,composed of industrial camera,lens,projector and computer,is effective,accurate and simple to use.The MATLAB and OpenCV are used to filter and preprocess the modulated raster image,then a series of parameters from the system calibration to combine together.Finally,the 3D reconstruction of the object is realized.