考虑自适应光学系统对波前处理计算量和实时性的要求, 对原有基于块运算的波前处理算法进行了改进和优化, 提出了一种自适应光学系统波前处理算法。该算法基于重复利用核心模块的方式完成波前斜率计算, 利用矩阵与向量相乘的可分解性完成波前复原计算。在像素时钟的同步下, 完成整个波前处理运算, 给出促动器所需的促动量。采用一片Virtex-4 LX80现场可编程门阵列(FPGA)作为核心处理芯片进行了实验, 结果表明该算法对于同样规模的自适应光学系统可降低约为50%的硬件资源, 提高了系统的波前处理规模; 与原算法帧同步结束后有338 μs的计算延时相比, 本文提出的算法可在帧同步结束前完成整个波前处理的运算, 提高了自适应光学系统的控制带宽。利用改进后的算法在原系统上进行了室内的光源校正实验, 取得了很好的效果。

In consideration of the requirements of an adaptive optical wave front processor for heavy computation and real-time capability, this article proposed an adaptive optical wave front algorithm based on block processing for improving and optimizing the original ones. The algorithm calculated the wave front slope by a reusing core module manner and complemented the wave front reconstruction with the decomposition of matrix-vector multiplication. In the pixel clock synchronization, the entire wave front processing calculation was implemented, then the actuator was given to promote the necessary momentum. A Virtex-4 LX80 Field Programmable Gate Array(FPGA) was as the core processing chip to perform an experiment and the results show that the algorithm for the same size of adaptive optic system can reduce the hardware resources about 50%, by which the system's wave front processing scale is reduced. As compared to the original algorithm of 338 μs time delay after the frame synchronization end, this proposed algorithm can complete the whole processing before the end of the frame synchronization, so that the system improves control bandwidths. The improved algorithm has been used in the original system to conduct a room light calibration experiment, and good results are achieved.