The ever-increasing number of vehicles on a global level signifies the need for communications between vehicles and the surrounding environment. Visible light communications (VLC) is a viable complementary technology to congested radio-frequency-based wireless systems. In order to increase the reliability of the VLC link, two novel algorithms based on (i) channel inversion (CI) and (ii) frame subtraction and CI (FSCI) schemes are proposed to successfully extract the data in a non-line-of-sight multiple-input–multiple-output spatial-division multiplexing optical camera communications system. We have adopted differential modulation and frame subtraction schemes and proposed a unique packet structure to mark the packet and the position of the footprint of transmitters (Txs) in the image frame. We show that the FSCI scheme with much simpler receiver structures can offer almost the same bit error rate performance compared with the hybrid selection/equal gain combining (HS/EGC) technique at lower transmit power (illumination) levels of <13dBm for a single Tx and improved performance at higher illumination levels of >20dBm for multiple Txs. Compared with HS/EGC, CI schemes have a higher tolerance to the spacing between Txs, where the payload threshold level can be set to a fixed value of 0.5.