3D Depth Camera Design Using TOF Area Array Sensor And Binocular Vision

Three dimensional(3D) Depth Camera has various applications in different fields such as smart surveillance system,automatic driving,obstacle avoidance,3D scene reconstruction,AR and etc.There are two methods to capture depth data for 3D Depth Camera.The first one is active depth measurement and the second is passive depth measurement.Active depth measurement can obtain the depth of scene in real time through the flight time from TOF(Time of flight)array sensor's emitted modulated infrared light to its reflected one.And this method will not be affected by low texture and repetition texture.However,the measurement is not very sensitive to the high texture area and will be influenced by the ambient light easily,the object's material and reflectivity.Passive depth measurement generally uses the technology of binocular stereoscopic vision.The depth of the measured image is not affected by high texture,ambient light.But in the areas of low,repeat and occlusion texture,there will exist wrong matching.The depth images have a poor effect.Therefore,the combination of these two methods will become a hot research topic for 3D depth camera.In this paper,through the analysis of the advantages and disadvantages of the two techniques above,we designed an effective algorithm to fuse the TOF depth map and binocular stereo matching depth map.The algorithm based on TOF depth image noise reduction processing was to improve its reliability.The TOF depth map of the low resolution would be returned to high resolution,and then mapped to the binocular image matching points.Finally we would get a high quality disparity map in the algorithm.Based on the above algorithm,a set of hardware device was designed to accelerate the real-time operation of the algorithm.This device used the Zynq-7000 series FPGA chip of Xilinx as the core processor.The integration of TOF depth map and binocular depth map were completed by the co-design of hardware and software.This would improve the performance of real-time and environmental adaptability in 3D depth camera.This paper firstly introduced the background and development status of 3D camera application,TOF depth image and combination technology of binocular stereo vision.Then the advantages and disadvantages of the TOF technology and the principle of binocular stereo vision were analyzed.Secondly,this paper introduced the principle of combination algorithm.Finally,the experiment data and results were analyzed in detail.The results also verified the innovation and superiority of this combination system.