Point cloud registration is a critical step in 3D reconstruction. In order to solve the problems of low speed, many iterations and low accuracy of the traditional Iterative Closest Point (ICP) point cloud registration algorithm, this paper builds an imaging system that consists of a 3D camera and a RGB module and proposes a new method that combines the Accelerated KAZE (AKAZE) algorithm with the Generalized Iterative Closest Point (GICP) algorithm. In this method, the AKAZE algorithm was used to match the feature points of RGB image, and the RGB image feature points were mapped to the corresponding point cloud data. Moreover, the GICP algorithm was then used to achieve the point cloud registration. Results show that, compared with the usual ICP algorithm, the fusion algorithm in this paper reduces the number of iterations, the average time is shortened by 41.29%, the time efficiency is greatly improved, and the registration effect is also significantly improved. The point cloud registration method proposed in this paper effectively solves the problem of low time efficiency of traditional registration method.