Abstract: At present, most point cloud registration algorithms are based on the geometric features of point cloud data for description. With the emergence of scanning devices that can simultaneously collect object coordinates and colors, in order to better utilize color information and study the description of color information in color point clouds, a 3DLGOP feature descriptor was proposed based on color distribution, fused with geometric feature descriptors FPFH and color feature descriptors CSHOT to design a mixed descriptor of FPFH-3DLGOP. Using a nearest neighbor ratio method to obtain the initial correspondence, random sampling consistency to removes the wrong correspondence and singular value decomposition (SVD) to obtain the 3D rigid body transformation matrix, a point cloud registration was completed for the matching relationship. The experiment results show that, utilizing the color and geometric features of point cloud data, the proposed feature descriptor can not only complete the registration of color point clouds fully, but also improve the matching rate and accuracy of registration.