In view of the registration between LiDAR data and aerial image fusion, this paper presents a registration method combining surface feature and point feature. Firstly, the depth image is generated by LiDAR point cloud, and the surface feature is extracted from the depth image and aerial image. On the basis of this, the SIFT operator is used to extract the point features for the registration of LiDAR point cloud and the aerial image. This method adopts a registration strategy from the surface features to the SIFT features, reducing the amount of data for registration of features and the computational complexity of SIFT algorithm. In this paper, three sets of data are selected from the data set provided by ISPRS.The experimental results show that this method can effectively reduce the number of feature descriptors of SIFT operator and the time of finding the correct matching point, and improve the registration efficiency under condition of ensuring the registration accuracy. This is applicable to LiDAR point cloud and aerial image registration including large area of feature area in urban area.