In this paper, a compressed sensing (CS) based framework of multi-target cooperative visible light positioning (VLP) is formulated to realize simultaneous highaccuracy localization of multiple targets. The light emitting diodes (LEDs) intended for illumination are utilized to locate multiple target mobile terminals equipped with photodetectors. The indoor area can be divided into a two-dimensional grid of discrete points, and the targets are located in only a few grid points, which has a sparse property. Thus, the multitarget localization problem can be transferred into a sparse recovery problem. Specifically, a CS-based framework is formulated exploiting the superposition of the received visible light signals at the multiple targets to be located via intertarget cooperation. Then it can be efficiently resolved using CS-based algorithms. Moreover, inter-anchor cooperation is introduced to the CS-based framework by the crosscorrelation between the signals corresponding to different LEDs, i.e., anchors, which further improves the localization accuracy. Enabled by the proposed CS-based framework and the devised cooperation mechanism, the proposed scheme can simultaneously locate multiple targets with high precision and low computational complexity. Simulation results show that the proposed schemes can achieve centimeter-level multitarget positioning with sub-meter accuracy, which outperforms existing benchmark schemes.