基于梯度滤波器的多尺度高精度微运动测量

Multiscale Approach Based on Gradient Filters for Micro-Motion Measurement with High Accuracy

微运动测量在微纳制造和生物医学等高技术领域担任了一个重要的角色.本文将梯度滤波器方法和多尺度方法相结合,提出了一种用于微运动测量的基于梯度滤波器的多尺度方法.在该方法中,利用多尺度金字塔迭代方法,对测量图像进行降采样和低通滤波,将较大像素的图像运动转化为多个小像素的图像运动进行估计,从而提高运动估计精度.提出的方法用于测量两个像素附近的M EM S微机械图像运动时,测量偏差达到了0.005个像素.模拟实验结果表明,这种基于梯度滤波器的多尺度方法能够实现高精度的微运动测量.

Micro-motion measurement plays a significant role in the high technology fields sucn as mlcro/nano manufacturing and biomedicine. A multiscale approach based on gradient filters for micro-motion measurement was proposed. The measured images were downsampled and lowpass filtered using the multiscale pyramid iterative approach, and the larger-pixel motion was converted into multiple small motions to estimate. Therefore, the estimating accuracy can be improved. For MEMS images, the measurement bias of the proposed method reaches 0.005 pixels for the motions near 2 pixels. Experimental simulations show that the multiscale approach based on gradient filters for micro-motion measurement can measure micro-motions with high accuracy.