单光子雪崩二极管(SPAD)阵列具有单光子灵敏度和皮秒高时间分辨率等优点，在三维成像、自主导航以及地形测绘等领域具有很大应用前景。然而，由于生产工艺限制，目前的SPAD阵列还存在像素规模比较小、各像素性能差异很大的缺点。直接采用具有这样硬件缺陷的SPAD阵列进行深度成像所形成的三维图像横向空间分辨太低，且图像噪声很大。此外，单光子灵敏度的SPAD很容易受噪声光子干扰而加剧图像质量劣化。针对这些问题，本项目拟从噪声滤除和数据融合两个角度出发，探索把彩色CCD相机的强度数据与SPAD阵列的飞行时间数据进行数据融合实现高质量、高分辨率三维成像的原理，研究基于自然场景纵向稀疏特性的噪声光子滤除方法，以及融合彩色强度数据的正则化深度图像高分辨重构方法，建立CCD相机辅助的SPAD阵列深度成像系统并开展验证实验。项目研究将提升SPAD阵列深度成像系统的工作性能，具有重要研究和应用价值。

The single-photon avalanche diode (SPAD) array has the advantages of single photon sensitivity and picosecond high time resolution, and has wild application prospects in the fields of 3D imaging, autonomous navigation and terrain mapping. However, the current SPAD array still has the disadvantages of small pixel format and pixel-to-pixel variations of SPAD properties due to production process limitations. The transverse spatial resolution of the three-dimensional image formed by depth imaging directly using the SPAD array with such hardware defects is too low, and the image noise is very large. In addition, the SPAD with single photon sensitivity is easily interfered by noise photons to exacerbate image quality degradation. Aiming at these problems, this project intends to start from two perspectives of noise filtering and data fusion, explore the principle of data fusion of the intensity data of the color CCD camera and the time-of-flight (TOF) data of the SPAD array to achieve high-quality, high-resolution 3D imaging. Study the noise photon filtering methods based on the longitudinal sparse characteristics of natural scenes, and the high-resolution regularization reconstruction method for depth image, with color intensity information fused. Establish a CCD camera-assisted SPAD array depth imaging system and carry out verification experiments. The project research will improve the performance of the SPAD array depth imaging system, which has important research and application value.