In this paper, we have attempted to classify necessary feature points for building cognition using adaptive splice critical more than once to improve the efficiency of building recognition technique on the level of cognition and matching. Due to the rapid change in urban environment, it is necessary to reflect the change in building by recognizing significant feature points of database. First of all, the feature points were extracted primarily through scale invariant feature transform (SIFT) and then eliminated falsely matched feature points. The random sample consensus (RANSAC) was applied to classify these points in the hidden area.

SIFT, RANSAC, occlusion region, feature point.