Visible light communication (VLC) is one of the significant communication research topics and 2D barcoding is one of the common applications of camera-based VLC. As the ubiquity of portable electronic gadgets has increased, barcodes no longer appear in its printed version only and they appear on the electronic display devices frequently. They are widely used in mobile commerce and social media applications nowadays. However, an accurate display-capture channel model applicable to a wide range of display and camera devices is still lacking. This channel modeling problem is challenging because noises, spatial and non-linear distortions, and the pixelated effect are introduced to the images captured by the cameras during the displaying and photo capturing process. Without an accurate chann...[ Read more ]

Visible light communication (VLC) is one of the significant communication research topics and 2D barcoding is one of the common applications of camera-based VLC. As the ubiquity of portable electronic gadgets has increased, barcodes no longer appear in its printed version only and they appear on the electronic display devices frequently. They are widely used in mobile commerce and social media applications nowadays. However, an accurate display-capture channel model applicable to a wide range of display and camera devices is still lacking. This channel modeling problem is challenging because noises, spatial and non-linear distortions, and the pixelated effect are introduced to the images captured by the cameras during the displaying and photo capturing process. Without an accurate channel model, the results of barcode decoding are often unreliable. To combat these issues, we propose a generic display-capture channel model and its associated parameter estimation scheme, thus allowing various channel impairments to be represented and eliminated. Non-uniform modulation is utilized to improve the reliability. To boost data density using the proposed channel model, an 8-level grey scale 2D barcode with a maximum spatial efficiency of 1 × 1 display pixel per module (or modulation symbol) has been developed. Its training overhead required for channel estimation only takes up 2.6% of the total number of modules and the training symbols are distributed in the barcode to increase the robustness. The channel model is then extended to represent the RGB color channels in order to cope with cross color channel interferences in a color barcode system. With the proposed accurate color channel model, the data density can be further increased by up to 3 times by utilizing multilevel modulation in RGB color channels.