Aiming at solving the problems that the structural parameters of disc insulators are still measured manually with low measurement efficiency and precision, a fast and high precision random speckle 3D reconstruction method based on GPU acceleration for insulators is proposed. This method uses random speckle projection to increase the texture details of the measured object, and proposes a high-precision disparity matching algorithm-a semi-global stereo matching algorithm based on reverse synthetic Gauss-Newton iteration (ICGN-SGM) to obtain dense and high-precision disparity data, and then based on the parallel computing architecture CUDA performs parallel acceleration of the matching algorithm to improve computing efficiency. Experimental results show that the proposed method can accurately, densely and completely restore the three-dimensional shape of insulators, the reconstruction accuracy is better than 0.06mm, and the reconstruction speed can reach 5 fps.

暂无