快速测定天体位置是天体测量学的主要研究内容之一。随着科学技术的发展进步，一方面由于CCD相机已经广泛的应用于地面天体测量仪器，所使用的CCD芯片的尺寸越来越大，相机的帧频也越来越快，使得单帧图像所包含的信息变得越来越大；另一方面，很多地面天体测量仪器需要长时间不间断的观测，所获取资料的积累是非常惊人的。因此，在天体测量仪器的观测过程中会产生大量的观测资料。如何对所获取的光学图像资料进行快速高精度的处理一直是天文学家非常看重的问题。采用GPU和CUDA作为天文光学图像资料处理的构架，较之于以往的CPU处理，在处理速度和实时性方面的有着明显的优势。随着CUDA6.0的出现，支持CPU与GPU统一寻址，也部分降低了应用软件的实现难度。

One of the main research contents in astrometry is quickly measured positions of celestial bodies. With the development of science and technology, on the one hand, CCD camera has been widely used in ground-based astrometry instrument, the size of the CCD chip is bigger and bigger, and faster and faster frame rate of the camera, everything makes the information contained in the single-frame image getting bigger; On the other hand, many ground-based astrometry instruments require long uninterrupted observations, the accumulation of acquired data is very alarming. Thus, during observation with astrometry instrument will acquire large amounts of data. How to process these optical images in as little time as possible is an important issue for astronomers. Observation data are processed based GPU and CUDA other than CPU, with obvious advantages in processing speed and real-time. With the appearing of CUDA6.0, the CPU and GPU can unified addressing, greatly reducing the difficulty of application software programming.