In this article, we propose a distributed algorithm for cooperatively pursuing an adversarial evader in an unbounded environment with cluttered obstacles. The algorithm relies on constructing the buffered evader-centered bounded Voronoi cell (B-ECBVC) in real time for each pursuer to safely chase the evader among obstacles. Based on the B-ECBVC, an encirclement control law and a capture strategy are proposed. Specifically, the control law drives each pursuer toward the centroid of its B-ECBVC to trap the evader, while the capture strategy guides it to reduce the distance between a team of pursuers and the evader by adaptively compressing the B-ECBVC. By integrating the control law and the capture strategy, the pursuers can rapidly approach the evader, while simultaneously maintaining the encirclement. To guarantee collision avoidance, a rapid and reliable approach for creating secure regions of pursuers by integrating separating hyperplanes and buffered terms into B-ECBVCs. In addition, the proposed pursuit method is further extended to a higher order dynamics system with avoiding moving obstacles. Our B-ECBVC approach is validated with various escape policies of the evader in dense obstacle environments. Moreover, real-time experiments with an autonomous evader and a human evader are implemented in a multiple mobile robot platform to validate the effectiveness of our approach.