Abstract: To accelerate the optimization of long zigzag paths for unmanned aerial vehicles (UAVs) in complex obstacle environments, we propose a DynamicEllipse-RRT^* (DE-RRT^*) path planning algorithm, which incorporates dynamic elliptical domain sampling. First, to mitigate local irregular turns caused by high randomness in sampling during path planning, we employ a sliding window method to adjust and segment the path, clarifying the main direction. Then, we introduce random breakpoints and an adaptive iterative threshold strategy to construct small elliptical domains for targeted sub-segment optimization. We dynamically adjust the number of iterations based on sub-segment length to conserve computational resources. Meanwhile, we utilize elliptical focal points as heuristic information to guide probabilistic direct-push growth points, accelerating the search for more optimal paths. Additionally, redundant path nodes are eliminated, and path corners are smoothed, producing paths more suited for UAV applications. Finally, we evaluate the proposed method in complex 2D environments with multiple turns and dense obstacles. Comparative results demonstrate that the DE-RRT^* algorithm finds better paths in the same or shorter time under the same environmental and parameter configurations than existing methods.