Abstract: In order to improve the ability of unmanned aerial vehicle (UAV) to detect obstacles, a fast edge detection algorithm is studied and an improved local binary pattern (I-LBP) operator is proposed to enhance the edge extraction effect. In HSV (Hue, Saturation, Value) color space, the I-LBP operator is used to describe the local texture features of pixels. The Hausdorff distance is used to confirm the edge pixels. And the contour of the obstacle is framed in the rectangular box. The algorithm improves the traditional LBP operator according to the fuzzy set theory. The simple binary description of the local texture features is extended to three-dimensional vector. It enhances the LBP operator’s ability to describe the local texture features and is noise-robust.The simulation verification is carried out with MATLAB. The results show that the I-LBP operator has good robustness and can identify obstacles under the condition of poor illumination and noise pollution. It also has high real-time ability, which can meet the requests of UAV in avoiding obstacles.

Key words: unmanned aerial vehicle (UAV) obstacle avoidance; HSV color space; improved local binary pattern (I-LBP) operator; fuzzy set theory; Hausdorff distance