The frictional dynamics of microball bearings needs to be understood for effective design and control of micromachines using such bearings. In the prior work by the authors and coworkers, a vision-based, non-intrusive measurement method was developed for characterizing the microtribological behavior of linear microball bearings. In this paper a novel scheme is proposed for computing the frictional force and the relative velocity between bearing elements using noisy position information extracted from the video. It enables one to reveal essential frictional characteristics despite the low capture rate (30 frames/second) of the camera. Experimental and numerical results show that the dynamic friction behavior is captured by a memoryless viscous model.