Active Stereovision based Real-time Detection of Deformation on a Rotating Wind Turbine Blade

Abstract

In order to detect the deformations on a rotating wind turbine blade, a real-time deformation detection method based on active stereovision is proposed. It casts laser pattern on the target blade in the bright outdoor ambient light condition to improve the performance of stereo matching. The light pattern is also used to help distinguish the foreground and background objects. The fast sum of absolute difference criterion is adopted for quick stereo matching. Then the combination of epipolar constraint, the uniqueness constraint, the ordering constraint, the continuity constraint, and the disparity range constraint is applied in the process of disparity optimization. Subsequently the reference plane is generated by surface fitting based on the point cloud aquired. At last the 3D points with abrupted depth changes compared with the reference plane are deemed as the deformation area and the width and depth of this area are calculated. Experiment results indicate that the proposed method is efficient and reliable in the real-time detection of deformation on an operating wind turbine blade.