Background removal using Gaussian mixture model for optical camera communications

Optical camera communication (OCC) is a special type of visible light communication (VLC) using complementary-metal-oxide-semiconductor (CMOS) camera as receivers. However, OCC is limited by the heterogeneous reflective background in non-line-of-sight (NLOS). A background removal method is presented in this paper based on Gaussian mixture model (GMM). GMM is used to model each pixel of the background frame, and the bright and dark fringes generated by the rolling shutter effect are obtained by subtracting the background frame from the data frame. Bit error rate (BER) performance with purpose method is given at different ambient illuminances and transmission distances. The BER is 9.3 x 10-4 under the condition of 350lux ambient light and 1 m distance with transmission rate of 1.2kbits/s.