基于LED阵列单光子探测的高分辨率实时三维成像方法研究

Three dimensional (3D) imaging can adequately describe the relative position and motion of the different objects in a real 3D scene, which has wide application in the field of machine vision, reverse engineering, automatic navigation, virtual reality and man-machine interaction. The project will study the principle, method and key technique on a LED-array-based high-resolution real-time 3D imaging sensor using single-photon detection technology. By using high sensitive Geiger-mode APD array detector and fast time-division electrical scanning close packaged two dimensional (2D) LED array TOF ranging system, the imaging sensor can obtain high-pixel, high-resolution and real-time range images without any mechanic moving part. Moreover, the imaging sensor also can achieve high precision registration and data fusion between range image and 2D image by sharing one lens method and using highly efficient interpolation algorithm. Last, a range imaging signal-to-noise ratio (SNR) model, which can illustrate the interaction mechanism between different parameters of imaging sensor, will be built based on the basic photoelectric ranging theory for guiding the optimized design of this imaging sensor. By studying on this project, it can overcome many deficiencies of those traditional 3D imaging technology, such as low detection sensitivity, low measurement speed and low registration accuracy. The anticipated results of the project will provide theoretical basis and technical support for developing new 3D imaging method and building high-resolution real-time 3D imaging sensor with small volume and low cost.

三维成像能够更加真实地描述三维场景中物体的位置和运动信息，在机器视觉、逆向工程、自主导航、虚拟现实、人机交互等领域具有广泛的应用前景。本项目开展一种基于LED阵列单光子探测的三维成像原理、方法及关键技术研究，拟利用盖革模式下APD阵列高灵敏的单光子探测特性，通过快速时分电扫描密集封装的二维LED阵列TOF测距系统，实现非机械扫描、高像素、高分辨率、快速的深度图像获取；通过探索共用透镜成像方法和高效的插值处理算法，实现深度图像与二维图像的实时、高精度配准与融合；同时开展光电测距理论研究，建立测距成像信噪比模型，阐明系统各参数间的相互作用机制，并实现成像系统的优化设计。通过开展本项目研究能够解决现有三维成像技术存在的探测灵敏度低、获取速度慢以及配准精度差等不足，其预期成果将为发展新型三维测距成像方法，构建小体积、低成本、高分辨率、实时的三维成像系统提供理论基础与技术支撑。

三维成像能够真实地描述三维场景中目标的相对位置和运动信息，在机器视觉、逆向工程、自主导航、虚拟现实、人机交互等领域具有广泛的应用前景。在项目执行的三年内，我们发展了一种基于LED阵列照明和时间相关单光子计数的三维测距成像新方法，先后探索了盖革工作模式下APD的高灵敏度单光子接收特性、研究了基于TDC的高精度飞行时间检测方法、建立了微弱散射光的单光子探测理论模型，并搭建了低像素的LED阵列三维成像原理样机，在户外50m远距离处，可实现采样时间优于1ms，且精度高于1cm的三维测距成像，同时还具有较强的抗环境光干扰能力；此外，该系统还被用于水中颗粒物散射目标的高灵敏度探测。相关研究结果先后被国外经典期刊IEEE Trans. Instrum. Meas., Opt. and Laser Tech.及OSA、SPIE会议等报道，其中论文被SCI收录2篇，EI收录4篇，先后申请国内发明专利3项，申请PCT国际发明专利1项，目前已授权2项。本项目的研究成果可为未来发展新型三维测距成像方法，构建高性能的三维成像系统提供理论基础与技术支撑。