A multiscale optimal filter method for micro-motion measurement with high accuracy

Micro-motion measurement plays an important role in technologies such as micro/nanomanufacturing and biomedicine. In this paper, micro-motion measurement is viewed as a signal processing problem, and the measured image gradients are estimated using the filter methods. A class of optimal filters for image gradient calculation is designed according to Parks-McClellan algorithm. In combination with multiscale approach, a multiscale optimal filter method for micro-motion measurement is proposed. In such a method, the larger motions are converted into multiple small motions to measure, thus the measurement accuracy can be further improved. The maximal bias magnitude of this proposed method reached 0.0064 pixels for the motions near 2 pixels. Experimental simulations show this proposed multiscale optimal filter method can measure the micro-motion with high accuracy.