To address the problem of cooperative perception for multiple intelligent and connected vehicles (ICVs) in road scenarios, this article proposes a cooperative perception registration algorithm for ICVs, which is based on sparse semantic features. The proposed algorithm aims to extend the perception range of ICVs by point cloud ensemble registration. Therefore, the cooperative perception for ICVs is achieved. Firstly, the sparse semantic features are obtained by geometric feature extraction based on road semantic features. Secondly, the angle deviation among the road semantic features is calculated to provide the initial registration value. The point cloud semantic information is used as the registration constraint condition to realize the global semantic ensemble registration. The experiments show that the proposed algorithm effectively expands the cooperative sensing range of multi-ICVs. The accuracy and robustness of multipoint cloud ensemble registration are enhanced. Compared with the current mainstream algorithm JRMPC, the registration accuracy of the proposed algorithm is improved by 2. 45% .