Structured Light 3D Camera System Desig

Structured light three-dimensional reconstruction technology is becoming more and more mature in terms of speed and accuracy,and is developing toward embedded integration while meeting the premise of high speed and high accuracy.At present,structured light 3D reconstruction system mainly relies on GPU(Graphics Processing Unit)to complete the calculation work of structured light decoding and three-dimensional coordinate mapping.However,the high power consumption and large size of GPU cannot meet the industrial demand of embedded integration.The FPGA architecture,on the other hand,can achieve low power consumption while playing an excellent computing power due to its configurable hardware resources and the advantage of fully parallel and pipelined computing.At the same time,the heterogeneous platform combining FPGA and ARM processor provides a new direction for the embeddedness of structured light 3D reconstruction system.Therefore,the adoption of FPGA architecture to accelerate the computation of structured light 3D reconstruction system has become a new trend in the industry technology development.This thesis designs a structured light 3D camera system based on ZYNQ,builds a phase method system model,and adopts the phase shift method to obtain the phase of the object so as to map the 3D coordinates,in which the work on the FPGA side is divided into image acquisition,cache transfer,phase calculation and 3D mapping,and the ARM side is mainly responsible for the communication and control,giving full play to the advantages of FPGA + ARM architecture software and hardware working together.The whole system is calibrated in MATLAB,the intermediate parameters needed for 3D mapping are obtained,and finally a complete phase-shifted structured light system is built,and 3D reconstruction simulation verification and experimental analysis are carried out.The specific work is as follows:1.The system model of solving absolute phase based on phase shift method is studied;the phase calculation module based on FPGA architecture is studied to improve the speed of phase calculation.The parallel input processing of 5-way phase-shifted stripe images is realized,while the bit width is controlled to reduce the hardware resource consumption.2.Researched the system calibration based on the phase method model,proposed a simple and fast calibration method to obtain the intermediate parameters required in the 3D mapping formula,solved the problem of high time consuming system calibration,and improved the accuracy of 3D reconstruction.3.The driving control of the camera and structured light projection module at the PS(Processing System)side of ZYNQ is completed;the Xilinx DDR3-based multi-frame image cache architecture is studied for the need of multi-frame image calculation by five-step phase shift method;the BRAM-based calibration data cache reading module and the data transfer communication module between the PL(Programmable Logic)side and the PS side is designed.4.The structured light systems of PC platform and ZYNQ platform were built respectively,and 3D reconstruction experiments were conducted and the results were analyzed.Finally,the 3D coordinates output from this system were displayed in point clouds,and the whole system was evaluated in terms of system resource consumption,reconstruction speed and accuracy.The experimental results show that even the ZYNQ xc7z020clg400-2 platform with less resources can fully satisfy the hardware consumption of the system,the reconstruction speed can reach 357 ms,which is 20 times of the reconstruction speed of the PC platform,and the reconstruction accuracy can reach 0.05 mm,which satisfies the measurement accuracy of phase shift structured light 3D reconstruction.