This work presents the implementation of a computer vision-based navigation system for trajectory tracking applied to an unmanned aerial vehicle. A system capable of interpreting and following trajectories in horizontal and vertical planes was developed using the DJI Tello drone and implemented in PyCharm through image processing and visual tracking algorithms. Processing utilized the onboard camera, applying segmentation in the HSV color space. For horizontal tracking, an algorithm based on virtual sensors was implemented that divides the image into three regions, determining translation or rotation actions for the drone. In vertical tracking, a state machine with virtual sensors distributed in upper and lower strips was used, enabling decisions based on the detection area. Experimental tests were conducted applying metrics such as total travel time, proximity percentage, average error, errors greater than 100 pixels, lateral corrections, estimated distance, and phase changes. Results demonstrate optimal system behavior, where horizontal trajectories achieved 91.67% closeness in smooth curves and 83.88% in tight curves. In vertical tracking, square and circular trajectories showed stability with completion times of 22.65 s and 22.86 s, respectively. However, the triangular trajectory presented greater instability with a mean time of 11.69 s and standard deviation of 2.98 s, indicating difficulties with trajectories containing acute angles.