A Novel Approach for Integrated Optical Camera Communication and Sensing Using Reflected Light

Optical camera communication (OCC) has gained much attention as it leverages the COTS cameras as receivers. While existing work mainly focus on direct light OCC, little has been done for refelcted light for OCC wiht complex background. As such, this paper studies the problem of OCC using reflected light by decoding the striped images captured by COTS devices and its coupled problem of background image reconstruction for scence sensing, and accordingly present a novel approach for integrated optical camera communication and sensing. To this end, we treat the decoding of images with dark and bright stipes as an object (i.e., dark or bright stripe) detection problem, and propose to modify the loss function of the state-of-the-art object detection model, i.e., YOLO V8, such that the width loss becomes more important. Based on the detected dark and bright stripes, we present an improved GAN model to reconstruct the background image by introducing dual-attention mechanism for feature fusion. The experiments show that our approach can accurately detect and decode the stripes in the striped images, and reconstruct the background images.