The unmanned aerial vehicle (UAV) offers unique advantages of autonomous flight capability and small coefficient of risk, and is increasingly being used in harbor-border inspection to ensure security and orderly operation of harbors. In response to the influence of external factors such as electromagnetic interference in harbor-border inspection, this paper utilizes UAV and visible light communication (VLC) to build an efficient system to track maritime targets near the harbor reliably. In a VLC scenario, a geometrical equation for transmitter positioning is first proposed based on the received signal strength of the optical signal emitted by the target. On this basis, linear iterative positioning (LIP) using first-order Taylor expansion is proposed to realize online beam tracking. Furthermore, quadratic approximative iterative positioning (QAIP), a more precise approximation of the geometrical equation, is proposed based on second-order Taylor expansion. Simulation results show that the proposed algorithms can track targets effectively, and QAIP can achieve higher accuracy with no noise or high signal-to-noise ratio. In addition, compared with the geometrical solution, LIP and QAIP have faster computing speeds and fixed overheads.