An Approach for Utility Pole Recognition in Real Conditions

In this work, we propose an approach for utility pole recognition in real conditions based on color, shape and photometric stereo vision, by using conventional low cost cameras. This subsystem is part of an automatic path planning system for a maintenance robot, which repairs the cable connections in electrical poles. This method could be used in applications requiring object recognition in outdoor environments. The challenges facing this approach include extreme solar illumination, the confusion between telephone poles, cable TV, in columns of buildings, trees, street lights, and tilt between the groundand the pole. The experiments of this recognition system shows satisfactory results under different solar illuminations, different distances between the post and the cameras, different inclinations of pole with respect to the ground, occlusions of the pole and location of the utility pole from cameras system. Results were totally satisfactory with 100% effectiveness in a range of 5% to 95% with respect to the H component of the HSV scheme. The proposed method recognizes and locates utility poles with respect to the stereo vision system.