Aiming at the problem of large number of pilots and low efficiency and accuracy in existing channel estimation algorithms of relay-and-forward indoor visible light communication systems, a channel estimation method based on tensor mode noise compensation is proposed. Firstly, a pilot structure which is suitable for noise compensation for the received data is designed by making the best of the characteristics of the transmission data structure in visible light communication. Then, according to the framework of PARATUCK2 tensor decomposition, a communication system with noise model of this pilot mode is constructed. Finally, combined with the tensor decomposition method, a method of estimating the real noise with noise compensation of pilot frequency is designed, and all channel parameters are estimated. The simulation results show that the estimation algorithm based on tensor mode noise compensation can be applied to the relay-and-forward indoor visible light communication system, accelerate the iteration speed, and improve the estimation accuracy, which fully verifies the effectiveness and feasibility of the algorithm.