相位恢复算法在仿真与实验上的研究

相位恢复算法一直存在着精确度不高,收敛速度慢甚至停滞不前等问题。将基于光强传输方程(TIE)法与G-S迭代算法混合提高了相位恢复的精确度,梯度算法的提出加大了迭代步长,使得收敛速度加快。采用GS-TIE算法和振幅加成梯度算法分别从仿真和实验的角度去比较分析恢复的效果。通过对二维图像仿真得出,振幅加成梯度算法在收敛速度上是GSTIE迭代算法的3倍,精确度是GS-TIE迭代算法的10倍。从实验结果得知,GS-TIE恢复的相位清晰可见,轮廓明显,在边缘处过度均匀,而振幅加成梯度算法相对比较模糊,在轮廓边缘处过度不均匀,悬差较大。

Research on simulation and experiment of phase retrieval algorithm

The phase retrieval algorithm has always had some questions such as low precision, long convergence time and even stagnation. Based on the transport of intension equation (TIE)-Gerchberg-Saxton (GS) iteration hybrid algorithm improves the precision of phase retrieval, and gradient algorithm increases the iteration step and accelerates the convergence speed. This paper compares and analyses the GS-TIE algorithm and amplitude addition gradient algorithm retrieval effect from the perspective of simulation and experiment for GS-TIE algorithm and amplitude addition gradient algorithm. Based on the 2D images simulation for the comparison of two phase retrieval algorithms, it proves that the convergence speed of the algorithm is three times that of GS-TIE iteration algorithm, the accuracy of the algorithm is better than GS-TIE iteration algorithm, and the accuracy is ten times better than GS-TIE iteration algorithm. The experimental results show that GS-TIE algorithm retrieval phase is clearly visible, the contour is obvious, and it is excessively uniform at the edge, while the amplitude addition gradient algorithm is relatively fuzzy, and it is excessively uneven at the edge of the contour and has a large suspension.