Research On Some Technologies In Three-dimensional Measurement Based On Binary Pattern With Projector Defocusing

Three-dimensional measurement with fringe projection is a technique based on optical theory,which also combines with computer vision,image processing,digital signal processing and other disciplines,in order to achieve 3D shape measurement on the surface of object.However,for traditional 3D measurement with sinusoidal fringe projection,the hardware problems such as the frame rate limit or the Gamma distortion will usually hinder the measurement ability for the dynamic object.Recently,the 3D measurement system with projector defocusing breaks through the bottleneck of the above problems and becomes the mainstream technique.In this thesis,some novelties and research work are carried out to solve the key issues of the 3D measurement system with projector defocusing,which are outlined as follows:1.A novel method for binary pattern generation is proposed,which ensures the high quality of fringe patterns under different levels of defocusing so that the projector need not to be strictly controlled,and so high precision and strong robustness of measurement results can be guaranteed.A new method of designing binary patch is proposed based on symmetry and periodicity of the sinusoidal fringe pattern.Even though the new binary patch is designed with much smaller size,it can not only effectively describe the feature of the binary pattern but also improve the processing efficiency.Furthermore,a new method of multi-pixels'mutation is proposed to optimize the binary patch.The new method avoids the local optimal solutions of single-pixel's mutation and improves the sinusoidal property of binary fringes.Compared with traditional binary fringe generation method,experiments certify that the proposed method is robust and can always obtain high-quality phase.2.A novel point cloud interpolation method is proposed to solve the problem that the resolution of the obtained point cloud is restricted by the resolution of the hardware of the 3D measurement system.At first,the relationship between phase and point cloud is explained theoretically to analyze the feasibility of the proposed interpolation method.Then three image interpolation methods are introduced and compared.Finally,an interpolation algorithm for unwrapped phase and a solution to the case that there are holes in the phase are proposed respectively.Experiments verify that good quality of point cloud data can be gained with this method.3.A 3D measurement system with projector defocusing is set up with above innovative achievements.Besides,a heterogeneous parallel computing model CPU+GPU is introduced to accelerate the three-dimensional measurement algorithm so that real-time measurement can be achieved.Experiments are carried out to verify that the system gets the advantages of high accuracy,strong robustness and high resolution in order that it can be used for real-time measurement and display of dynamic object within a certain speed.