A receding horizon and topology optimization (RHTO)-based autonomous exploration method is proposed to address the issues of lack of foresight, large exploration areas, and susceptibility to ocean currents in multi-region autonomous exploration for unmanned surface vehicles (USVs). First, the overall exploration task is decomposed using a receding horizon approach, which estimates the optimal global exploration path for a future period based on currently known information and executes only a portion of the path. Second, a basic path network connecting currently accessible regions is rapidly constructed through frontier detection and cyclic sampling. This is combined with a topology optimization algorithm to quickly converge to the shortest path network. Finally, a cost function is used to evaluate the navigation cost of each path. A genetic algorithm is applied to plan the optimal exploration sequence and paths, generating coverage waypoints as viewpoints to guide the USV's exploration before eventually returning to the starting point. Simulation results demonstrate that, compared to several common autonomous exploration algorithms, the RHTO method achieves the smallest values in three key metrics: total path length, total path turning angle, and energy consumption to overcome water resistance, while also exhibiting better robustness.