Visual Tracking With Motion Distortion Removal for Nanomanipulation Inside SEM

The paper investigates the problem of visual tracking on moving object in nanomanipulation inside scanning electron microscope (SEM). Image noise is a primary concern when dealing with the problem, which includes the inherent statistical noise and that induced by the motion distortion. A visual tracking method with SEM image denoising algorithm is proposed. The denoising algorithm is well incorporated with robot motion by innovatively leveraging the image Jacobian matrix technique. The denosing algorithm removes image noise and provides more realistic image. On the basis, template matching is utilized to achieve visual tracking on image plane. Experimental results show that the visual tracking performance on image plane was well improved by 31.4% for point feature, 60.6% for line feature, and 52.1% for area feature in terms of root mean square (RMSE) compared to that without motion distortion removal. Comparison experiments validate the state-of-the-art performance achieved by the proposed method and thus the superiority.