Visible light communications (VLC) have been recently proposed to enhance the capacity of next generation of wireless services. Moreover, VLC networks usually comprise a large number of overlapping optical access points (APs). Moreover, each of these APs provides a small and confined area of coverage in order to generate satisfactory illumination. In this work, a user-centric (UC) clustering formation based on the K-means algorithm is proposed to manage the inter-cell interference (ICI) and enhance the performance of VLC networks. Moreover, assuming that each user is equipped with a reconfigurable photodetector, the use of blind interference alignment (BIA) in each UC cluster is considered. Notice that the data rate demands are not the same for all the users. We formulate an optimization problem to maximize the utility of the network resources allocated to the users based on their demands. After that, a centralized algorithm is proposed to obtain an optimal solution through exhaustive search, which is subject to high complexity. To reduce the complexity of this optimization problem, the problem is divided into sub-problems based on the number of constructed UC clusters. Then, a distributed algorithm via Lagrangian multipliers is proposed within each UC cluster with the aim of providing a near optimal solution to the centralized algorithm. Simulation results demonstrate that the proposed resource allocation algorithms provide higher performance than a uniform resource allocation scheme among users.