Research On Fringe Projection Profilometry Technique For Dynamic Measurement System

Fringe projection profilometry is a very popular non-contact 3D measurement technique at present,which is widely used in fields of industrial testing,reverse engineering,medical diagnosis,heritage conservation and so on because of its high accuracy,high speed,high resolution and high robustness.Current fringe projection profilometry has mature technique for measuring static objects.However,the speed and the accuracy are two main obstacles when the fringe projection profilometry is applied in dynamic measurement because of object motion.Until now,many phase unwrapping approaches improve the speed at the sacrifice of accuracy.And the nonlinearity error,the jump error,the motion error and the motion artifact are four main factors for restraining the measured accuracy.Based on these problems in dynamic measurement,this paper try to develop according solutions.At present,reducing the number of gratings is a method for improving the measured speed.Based on template matching and morphology,this paper presents a code-based phase unwrapping method.More than 36 fringe orders can be obtained accurately from only one code grating.The simulation and experiment demonstrate that the method has high accuracy and high robustness for measuring static scene and dynamic scene.Code-based phase unwrapping method has been widely studied since it uses fewer gratings to measure 3D objects.However,the fringe order obtained by code grating is misalignment with the wrapped phase due to the noise and defocusing.And the jump errors caused by the misalignment serious hinders the accuracy in the code-based approaches,especially applying in the binary defocusing projection technique.This paper summarizes the current solution ways and analyzes the relation between jump error and grating fringe,and then develops a self-alignment technique.Experiments show that the proposed technique can eliminate the jump error in complex-texture and step-height objects with high speed and no accuracy losing.In order to further improve the acquisition speed of 3D point cloud,this paper introduces dual camera and single projector system to increase the visual constraints,and recombines the speckle image,and combines with the binary defocusing projection technique.Based on the system,we design high-speed binocular matching function to improve the measured speed four times as the conventional single camera system.Since the binary defocusing projection technique has the ability of immunizing the non-linear error,the phase obtained by this system has no use for the revising of no-linear error.In addition,we use the current high-precision algorithm to compensate the motion error.And then an approach for eliminating motion artifact is proposed to improve the accuracy of 3D point cloud.Last,the highly dynamic bionic bird with high velocity(303fps)is reconstructed by this system.The above researches develop two high-speed measurement approaches based on single-camera and dual-camera.With respect to that non-linear error,this paper uses the binary defocusing projection technique to avoid.On the issue of jump error,this paper provides a self-alignment technique without accuracy loss.We use the Hilbert transform method to compensate the motion error,and a designed approach is proposed to eliminate motion artifact.