This work describes computer vision methods developed for autonomous recharging of a quadrotor from a standard wall outlet. Specifically, this work encompasses two algorithms for detecting and tracking an outlet in a video stream acquired by a low cost quadrotor platform. Two different algorithms were developed for assumptions and requirements associated with different distances between the quadrotor and an outlet. The close-range algorithm achieves higher processing frequency and greater accuracy, but is effective within a limited distance to the outlet. The longrange detection algorithm requires greater processing time, but imposes fewer assumptions and can detect outlets at greater distances (lower resolution). The algorithms overlap in their effective ranges and therefore allow the system to continuously track from over 6 feet to within several inches of an outlet. The close-range algorithm has achieves a processing frequency greater than 15 Hz. The long-range detector accomplishes scale invariant hole detection with automatic outlet scale selection.