Motivated by the directional sensing feature of video sensor, a direction adjustable sensing model is proposed first in this paper. Then, the coverage-enhancing problem in directional sensor networks is analyzed and defined. Moreover, a potential field based coverage-enhancing algorithm (PFCEA) is presented. By introducing the concept of “centroid”, the pending problem is translated into the centroid points’ uniform distribution problem. Centroid points repel each other to eliminate the sensing overlapping regions and coverage holes, thus enhance the whole coverage performance of the directional sensor network. A set of simulation results are performed to demonstrate the effectiveness of the proposed algorithm.