This research presents a novel variable baseline stereo vision system to overcome the limitations of traditional fixed baseline technology in complex environments. Traditional stereo vision systems struggle with large depth variations and visual dead zones due to their fixed baselines. In response to the scenario where UAVs are equipped with a single camera and take multiple images from different locations to simulate stereo vision, a modified algorithm is proposed. This algorithm adjusts the baseline according to the target’s distance and angle, improving depth resolution. Through improved calibration techniques and sensing systems, the algorithm accurately identifies the target’s 3D position. Experimental results show that, compared to traditional methods, the variable baseline system significantly enhances distance measurement accuracy in complex environments, demonstrating potential in applications such as mountain rescue and wildlife conservation. Future work will focus on enhancing the system's intelligence and expanding its application scope and practical value.