通过对微机械薄膜变形镜影响函数矩阵的奇异值分解, 构建了变形镜可以校正的像差模式空间, 分析了变形镜对正交基模式的校正能力和校正范围, 提出了一种变形镜闭环迭代控制算法。通过对影响函数矩阵低秩近似的方法滤除校正性能较差的基模式的影响, 实现了对畸变像差的有选择校正。以人眼出射波前为对象进行实验, 通过比较不同模式项数校正时的效果, 确定了最优校正模式项数的范围, 通过设置残差容限的方法, 消除了人眼晃动和眨眼的影响。实验结果表明:控制算法能通过选择合适的校正项数, 提高变形镜的校正性能, 并获取到了高清晰度的人眼视网膜图像, 为眼科疾病诊断和治疗提供了一种新的观察手段。

The correctable aberration modes space of the micromacined membrane deformable mirror(MMDM) is constructed through the singular value decomposition of the mirror’s influence function matrix. Analyzing the correction ability and range of MMDM for a set of orthogonal basis modes, a closed-loop iterative control algorithm for MMDM is introduced. By means of low-rank approximation of the influence function matrix, the algorithm can correct the aberration selectively by filtering out some aberration modes which can not be corrected easily. In experiments, the wavefront from a human eye is used as the incident wavefront of adaptive optics system. The range of the optimal correction mode number is determined by comparing the correction effects for human eye aberration with different numbers of modes, and the influence of eye shaking and blinking are eliminated by setting error tolerance of residual aberration. The results show that the control algorithm can improve the correction performance of MMDM significantly by selecting a suitable number of modes, and the high-resolution retina image can be obtained successfully. This paper provides a new observation means for ocular disease diagnosis and treatment.