The Research On Dynamic Scenes 3D Information Acquisition Based On Phase Measurement For Fringe Projection

Three-dimensional(3D)information acquisition technology has been widely applied in human modeling,pattern recognition,industrial testing,reverse engineering and target recognition,which shows the strong commercial prospect and the high research value.Phase measurement profilometry(PMP)based on fringe projection is one of the important 3D information acquisition techniques at present.As the applying fields of PMP extend,high density&quality 3D information acquisition for dynamic scenes using PMP is becoming one of the difficult and hot research fields.Around the key technology about dynamic scenes 3D information acquisition based on PMP,this thesis mainly includes several research aspects:the PMP technology on binary patterns with projector defocusing,the calibration method of the PMP system on binary patterns with projector defocusing,the research of the phase errors on binary patterns with incomplete defocusing of projector,the research of the interframe synchronization errors with the moving objects,and the PMP technology on complementary decoding with fusion framework of dual projection equipment(EVMs).Through the above research,this thesis provides the theoretical and experimental basis for the further establishment of high-speed PMP system.The major contents and research results of this thesis are summarized as follows:1.This thesis researched the PMP technology on binary patterns with projector defocusing.In the research,the generation methods of binary patterns based on spatial structure aware halftoning and the temporal-spatial combination were respectively proposed.For the realization of the first method above,the structure aware halftoning algorithm was introduced to construct the objective function and the optimization framework with the "patch" with the periodicity and symmetry of fringes was considered.For the realization of the second method above,the principle of linear superposition of PMP system was used as a basis and the hardware characteristics of the projection equipment and digital camera were considered.Finally,the proposed approaches were verified to achieve a high-quality near sinusoidal fringe patterns through the simulations and experiments.2.This thesis researched the calibration method of the PMP system on binary patterns with projector defocusing.In the research,the parameters of nonlinear camera model were optimized by the bundle adjustment algorithm.In addition,the calibration method with projector focusing,which was based on the light idea and the assist of double reference planes,was proposed.Then,the principle of phase domain invariant mapping was introduced and subpixel.matching method using iteration was proposed,and these were used as the basics for the proposed calibration method with projector focusing applying in the calibration method with projector defocusing.Finally,the accuracy and reconstruct ability with high reflective,colorful or complex objects of the proposed calibration approaches were verified by the experiments.3.The thesis analyzed the phase error on binary patterns with incomplete defocusing of projector and the interframe synchronization error of moving objects.Based on these,the error compensation methods were proposed.For the first error above,the theoretical derivation,simulations and experiments were adopted.Then,according to the experimental results,the two-dimensional(2D)fitting polynomial error model was introduced to compensate error by post-correction.For the second error above,the linear and quadratic models were yielded by theoretical derivation.According to the simulation results,2D fitting polynomial error model was proposed.Then,three kinds of error models were compared to illustrate the accuracy and adaptability of the proposed model.Based on the proposed error model,the Nelder-Mead algorithm was combined to realize error post-correction compensation.Finally,the effectiveness of the two error compensation methods was verified by experiments.4.The thesis researched the PMP technology on complementary decoding with fusion framework of dual-EVMs.In the research,the dual-EVMs fusion framework with the position relationships of strict-constraint,quasi-constraint and non-constraint between two projection devices(DMD).For the strict-constraint method,the DMDs of the upper EVM and lower EVM were set to strictly align,and the two-frequency phase unwrapping method of co-prime wavelength was utilized to realize the less capturing frames.For the quasi-constraint method,the DMDs of the upper EVM and lower EVM were allowed to slight misplace,and the complementary gray-code phase shifting method was utilized to realize the less capturing frames.For the non-constraint method,the positional relation between the two DMDs of the upper EVM and lower EVM had no special requirement,and three principles or methods were proposed,i.e.,the depth information acquisition principle of low-resolution based on periodic time-space coding,the phase unwrapping method aided by the low-resolution depth information and the calibration strategy based on the data fusion of double measurement systems.Similarly,using these proposed methods or principles could realize the less capturing frames.Finally,the reconstruct ability of the 3D information acquisition for complex or moving objects was verified by the experiments.