| This paper designs a depth imaging system based on the time of flight(TOF)principle,researches include distance imaging algorithm,depth imaging with multiple device,adaptive parameter control,depth imaging in motion scene and high dynamic scene and phase wrapping cancellation.We analyse system parameter and depth error,then propose a fast depth error calibration method.Depth imaging measures distance of a scene and outputs depth image.The TOF method has high precision in distance calculation and imaging is real-time.TOF imaging system can be used for object recognition,scene reconstruction and 3d interaction.The system realizes 12 m distance imaging,with maximum error of 5cm indoor and 10 cm outdoor,and supports usages of multiple device,motion scenes and outdoor scenes.In this paper,the main works include design of system hardware and software platform,researches of TOF depth imaging algorithms and optimization,design of high precision imaging strategies in a variety of environments,error analysis of TOF system,and propose an efficient error calibration method.The details are as follows.(1)Design and implement of hardware and software platforms.The image acquisition module was designed with EPC660.The main control platform was designed with RZ/A1 H ARM processor.For outdoor usage 940 nm infrared LED array is designed as light source,and interface between the minimum system circuit and peripheral devices was realized.Development of CEU and USB2.0 drivers to ensure high speed image transmission,I2 C drivers configure a variety of working modes.Implement of image sensor exposure,data readout and transmission process of correlation sampling images.Eventually,a reliable image acquisition and transmission platform is realized.(2)This paper researches the difference between ideal model and implementation of AMCW distance algorithm,proposes increasing correlation sample frames to effectively reduce the interference of harmonics on distance accuracy.A pseudo-random noise sequence(PN)modulated light source is used to solve the interference problem of multiple device usages.In order to ensure quality of distance imaging,parameters such as the return light amplitude,exposure time and background light are researched,and an adaptive exposure time control algorithm is proposed to ensure the accuracy and stability of usages in dynamic scenes.(3)Depth imaging models of motion scene are researched,and a single-frame double-sampling method is proposed to suppress occurrence of motion blur and ensure accuracy of depth imaging.Depth imaging characteristics of high dynamic scene are researched,we propose a separate exposure hardware design with amplitude and saturation detection method to optimizes the imaging quality in high dynamic scene.Aiming at phase wrapping problem of TOF,a method of phase unwrapping by using amplitude information and dual-frequency depth measurement is proposed to effectively enlarge the maximum measuring distance.(4)We analyze the parameters of the imaging system,such as warm-up time,exposure time and modulation frequency to optimize working state of the system.TOF system pixels error,fixed pattern noise and the fourth order harmonic errors are researched.In this paper,a new method is proposed to store error data using the look-up table,and the DLL hardware delay is used to replace multiple distance measurement,so as to improve the efficiency and accuracy of the calibration process.After error calibration,the system realized that the maximum distance error of 5cm indoor,and the 10 cm error in outdoor strong ambient light. |
