Among the methods proposed to detect extrasolar planets, microlensing is the only technique that can detect free-floating planets. Free-floating planets are detected through the channel of short-duration isolated lensing events. However, if a seemingly isolated planetary event is detected, it is difficult to firmly conclude that the event is caused by a free-floating planet because a wide-separation planet can also produce an isolated event. There were several methods proposed to break the degeneracy between the isolated planetary events produced by the free-floating and wide-separation planets, but they are incomplete. In this paper, we show that free-floating planets can be securely identified by conducting astrometric follow-up observations of isolated events to be detected in future photometric lensing surveys by using high-precision interferometers to be operated contemporarily with the photometric surveys. The method is based on the fact that astrometric lensing effect covers much longer range of the lens-source separation than the photometric effect. We demonstrate that several astrometric follow-up observations of isolated planetary events associated with source stars brighter than $V\sim 19$ by using the {\it Space Interferometry Mission} with an exposure time of $\lesssim 10 {\rm min}$ for each observation will make it possible to measure the centroid shift induced by primaries with projected separations up to $\sim 100 {\rm AU}$. Therefore, the proposed method is far more complete than previously proposed methods that are flawed by the limited applicability only to planets with projected separations $\lesssim 20 {\rm AU}$ or planets accompanied by bright primaries.

5 pages including 2 figures