The optical fringe code modulation and recognition algorithm based on visible light communication using convolutional neural network

Recently, visible light communication (VLC) based on complementary metal-oxide-semiconductor (CMOS) sensor has been widely studied, and most of the research uses the modulation and demodulation method, which modulates the light emitting diode (LED) light to transmit data and demodulates bright and dark stripes on the image captured by the CMOS sensor to get data. However, the method have some defects. Firstly, as the distance increases to a certain extent, the data frame structure will be partially lost. Secondly, the image captured by the CMOS sensor must be strictly synchronized, which is hard to guarantee. Thirdly, the focus of recent related research is mainly on the real-time nature of communication, which is difficult to achieve due to the complex image processing methods at this stage. What is more, for many application scenarios recognizing the LED information in the image captured by the CMOS sensor is enough. So, in this paper, we introduce the RGB-LED and propose an optical fringe code (LED-OFC) modulation and recognition algorithm based on VLC using convolutional neural network (CNN). The RGB-LED is modulated to assign different features to the LED-OFC instead of transmitting data. Then the CNN is employed to recognize LED-OFC with different features. The experiment results show that both the recognition accuracy, the recognition amount, the maximum recognition distance and the robustness are greatly improved by the proposed method compared with the traditional modulation and demodulation method, which has broad application prospects.