Abstract
We have been developing methods of identifying moving objects and persons having an accelerometer of interest from a camera view. With our previous methods, it has been empirically pointed out that synchronously moving objects are often misidentified. In this paper, we first examine this pointing out through experiments of discriminating two synchronously walking persons, one of whom has the accelerometer and the other does not. Then we apply the coherence function to accomplishing the same discrimination problem. The experimental results show that the coherence function has higher discriminative ability than our previous methods, although it requires longer time to yield stable results.
