基于光流技术和滤波器方法的多尺度微运动测量

Multiscale micro-motion measurement based on optical flow and filter method

将光流技术和多尺度方法相结合,提出了一种基于光流技术和梯度滤波器的多尺度高精度微运动测量方法。该方法将微运动测量看成是光流的计算,采用滤波器方法计算图像的梯度,并利用多尺度方法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。该方法用于测量2个像素附近的微陀螺图像运动时,最大测量偏差小于0.008个像素。实验模拟结果表明,这种采用光流技术和多尺度相结合的方法能够实现高精度的微运动测量。

A multiscale approach with high accuracy for micro-motion measurement based on optical flow and multiscale technique that incorporated the optical flow method into multiscale approach was proposed. In the method, the micro-motion measurement is viewed as the optical flow computation, and the image gradients were computed using the filter method. The larger pixels motions were converted into multiple small motions to estimate using the multiscale pyramid iterative approach. Therefore, the estimator accuracy can be improved. For the gyroscope images, the maximal measurement bias of the proposed method is smaller than 0.008 pixels for motions near 2 pixels. Experimental simulations show that the proposed approach can measure micro-motions with high accuracy.