Intelligent reflecting surface (IRS) is a promising technology that improves the performance of wireless communication systems by constructively adjusting the communication environment. In particular, IRS can provide spatial degrees of freedom by creating artificial paths, when the rank of the multiple-input-multiple-output (MIMO) channel matrix is deficient due to the line-of-sight (LOS) environment. In this paper, we consider a multiple IRSs-assisted MIMO system in the LOS environment. In the high-frequency bands, using a large number of antennas is essential. However, the prohibitive hardware cost and power consumption of a full digital system must be overcome. Therefore, to maximize the cost-effective potential of the IRS-assisted MIMO system, a low-cost design of the transceiver also should be considered. Motivated by this, we jointly design the passive beamforming of the IRS and the analog/digital hybrid beamforming, which can achieve the performance of the full digital beamformer while reducing the number of radio frequency chains. For this, a non-convex optimization problem is formulated, which is separated into sub-problems. First, IRS phase shifters matrices are obtained by solving the channel power maximization problem. After that, the problem of the minimization of the difference of the Frobenius norm of the full digital beamformer and the hybrid beamformer is solved sequentially. Simulation results show that the sum rate of the proposed scheme is close to that of the full digital system.