Visible light communication (VLC) combined with advanced illumination may be expected to become an integral part of next-generation heterogeneous networks. In order to mitigate the performance degradation imposed by the intercell-interference (ICI), a user-centric (UC) cluster formation technique employing vectored transmission (VT) is proposed for the VLC down-link system, where multiple users may be simultaneously supported by multiple access points (APs). In contrast to the traditional network-centric (NC) design, the UC-VT cluster formation is dynamically constructed and adjusted, rather than remaining static. Furthermore, we consider the critical issue of multiuser scheduling (MUS) relying on maximizing the “sum utility” of this system, which leads to a joint cluster formation and MUS problem. In order to find a practical solution, the original problem is reformulated as a maximum weighted matching (MWM) problem relying on a user-AP distance-based weight and then a low-complexity greedy algorithm is proposed, which offers a suboptimal yet compelling solution operating close to the optimal value found by the potentially excessive-complexity exhaustive search. Our simulation results demonstrate that the proposed greedy MUS algorithm combined with the UC-VT cluster formation is capable of providing an average user throughput of about 90% of the optimal throughput, which is about three times the throughput provided by the traditional cellular design in some of the scenarios considered.