Intensity estimation of LEDs from spatially and temporally mixed images in image sensor communication

We focus on the road-to-vehicle visible light communication that system is composed of LED matrix as transmitter and high-speed camera as receiver. In proposed method, there are two problems. First problem is in case of long distance. It is difficult to receive data correctly because neighboring LEDs on the received image are merged due to less number of pixels and/or blurring by defocus. In this paper, we propose estimation method of LED intensity from merged image. We evaluated the method by measuring Bit Error Rate and confirmed effectiveness of the method. We also performed real time communication and succeeded to extend communication distance from the previous work by applying the method. Second problem is that, the transmitting rate has been reduced to half of the frame rate of the camera, to avoid intersymbol interference because of exposure time of camera frame. We propose a method to improve the communication rate. The transmitting rate is set same to receiver´s frame rate, and the sent data is estimated from pattern mixed image. Using 8b/10b encoding instead of conventional reversal pattern to reduce the flickering of transmitter´s lighting is also proposed.