Barcolits: Barcodes using LED tags and optical camera communications

Abstract

This work presents Barcolits, active LED tags that exploit the sequential acquisition of rolling-shutter cameras to materialize virtual barcodes in images. The generated barcodes have valuable properties. Compared to physically printed bar-codes, its size (in pixels) does not change with the distance to the camera. Also, its stripes are always vertically aligned in the image independently of the rotation and position of the camera. Moreover, as Barcolits are active devices, they can be scanned under low light conditions or from reflections in walls or mirrors. Furthermore, this approach reuses conventional barcode detection and decoding algorithms, allowing interoperability and backwards compatibility with current barcode scanners (image or photodiode-based). These benefits have been experimentally validated using three different barcode encoding schemes: Code 128, EAN-13 and ITF. In the experiment conducted, these barcodes are compared in terms of their minimum required size to ensure high reliability, their data density, their signal level, and their likelihood to produce flickering in the LED source. The experiment results show that Code 128 are ideal for high throughput data sharing, with the highest achievable data density of 4.92 bits per pixel and the lowest power consumption. It outperforms standard manchester codes by encoding 42% more bits using the same payload size. On the other hand, EAN-13 barcodes require the smallest size for a successful scan but produce noticeable flickering if a long stream of different barcodes is transmitted. Finally, ITF codes mitigate flickering at the cost of reducing their density. This extensive analysis provides a comparison tool to help choose the optimal barcode scheme for different applications and environments.