A robust transform-domain LED-ID recognition method under imperfect camera exposure conditions

Optical camera communication(OCC), leveraging light emitting diode(LED) and complementary metal oxide semiconductor(CMOS) cameras, is increasingly being used for indoor positioning. Recognition of LED-ID is one of the most critical steps in such methods. The exposure effect of CMOS cameras will decrease the stripe contrast when the LED frequency exceeds the exposure frequency. The traditional binary demodulation method is easily interfered with in this frequency band, which may lead to recognition failure. This article proposes a robust LED-ID recognition method that combines discrete cosine transform (DCT) and high pass filtering (HPF). The method utilizes phase resonance and cancellation in the transform domain to separate high-frequency information from low-frequency interference, successfully increasing the recognition frequency band by at least 2.5 times. At the same time, the camera sampling parameters and exposure effects on LED-ID recognition are analyzed to guide hardware selection for visible light positioning systems. The experimental results validated the recognition performance of the method in low-contrast situations and demonstrated its practicality and robustness.