激光散斑的亚像素位移法计算及比较

根据激光散斑相关测量中采用的插值法、梯度算法和牛顿一拉夫森迭代法3种亚像素位移算法的特点，逐一在亚像素尺度计算出磁流体干燥过程中的同一激光散斑图的偏均值和标准差变化曲线。基于得到的偏差均值和标准差讨论了3种算法处理激光散斑图的特点，比较了3种算法处理散斑图的计算精度、稳定性和计算用时。结果显示，在小变形的亚像素位移测量时，梯度算法的精度较高，且较为稳定，计算效率也较牛顿一拉夫森算法快约9倍。因此，在进行小变形亚像素位移测量时，采用梯度算法计算激光散斑是有效的。

Computation and comparison of laser speckle with sub-pixel measurement methods

On the characteristics of the interpolation, gradient-based and Newton-Rapshon algorithms used in the sub-pixel process in the laser speckle technology, the same laser speckle pattern was calculated one by one with the three sub-pixel algorithms under the drying process of the magnet fluid, and the mean bias errors and the standard deviation curves in a sub-pixel scale were worked out. The characteristics of the three algorithms to process the laser speckle pattern given by the mean bias error and standard deviation were discussed and the calculation accuracy, stability and calculation consumption were compared. It is shown that the precision and stability of the gradient-based algorithm is higher in the small deformation within the sub-pixel displacement measurement. The computing efficiency of the gradient-based algorithm is approximately nine times faster than that of Newton-Rapshon method. Therefore, it is more efficient to take advantage of the gradient-based algo- rithm to calculate the laser speckle in the small deformation and the sub-pixel displacement measurement.