Obstacle Recognition under Low Illumination for Inspection Robot

Abstract

Obstacle recognition is remain a challenge especially for the inspection robot of high voltage transmission line under varying degrees of illumination environments. This paper proposes an effective method to overcome the influence of low illumination. The proposed method firstly uses homomorphic filtering to attenuate high frequency illumination component; it then applies the Hu moment to extract obstacle features. The extensive experiments show that the proposed method is superior to four state-of-the-art algorithms, and achieves 88%, 87%, 90% obstacles recognition accuracy rate on shockproof hammer, suspension clamp and insulator string databases, respectively. Compared with other peer algorithms, homomorphic filtering has strong anti-illumination capability and the algorithm based on moment feature has higher recognition rate. The research improved the robustness of vision system of line-patrolling robot to image recognition, and promotes the safe, efficient, intelligent and sustainable development of power industry.