In recent years, with the continuous exploitation of marine resources around the world, the problem of underwater data communication has gradually become a research hotspot. This paper establishes an optical domain transmitting imaging reception model and proposes a MIMO communication system based on it, which the received spot must be aligned with the detector arrays. Moreover, it designs a transmission system based on this model by FPGA to complete real-time signal processing. However, the physical movement, turbulence, and other causes of the transceiver link misalignment will make the light spot offset the detector target surface, thus reducing the receive signal-to-noise ratio (SNR) and limiting the transmission distance. So, this system uses optical fiber to couple the received optical signal to the detector arrays, which reduces the alignment requirements. Further, we propose a Lattice Reduction zero forcing (LLL-ZF) detection algorithm for MIMO signal detection, which can reduce interference effectively. This imaging reception mode and MIMO signal detection can reduce the mutual interference between sub-links and decrease the complexity of the detection algorithm. The experimental results show that when the system’s BER is less than the FEC threshold value at a communication distance of 30 m underwater, the sub-channel communication rate of the system can reach 20 Mbps and the 4 \(\times \) 4 MIMO system rate can reach 80 Mbps.