Daniel Taller: Detecting and Decoding Barcodes in Images

Detecting and Decoding Barcodes in Images

Daniel’s research focused on making more efficient and robust algorithms to localize and then decode barcodes within images. This should allow consumers to read barcodes via camera phone to quickly find product descriptions, reviews, and pricing information. While current methods to do this exist, they are often not reliable for dim, unfocused, noisy, or low-resolution image that most camera phones produce.

After an image is obtained, the first step is to localize the barcode, or to find its location within the image. Most methods to do this search for an extended area of the image where there are strong gradients in a single direction. Daniel successfully implemented several such algorithms, varying from simple techniques involving numerical differentiation to more complex ones which made use of signal processing.

Once the barcode had been found, it must be decoded in order to obtain the product’s information. This is a much more delicate process, requiring me to about precise estimates of each bar’s width. In order to do this, Daniel implemented developed several techniques, which made use of statistical methods to identify the locations of the bars and mitigate noise and blur effects.