Blind Deconvolution of Barcodes via Dictionary Analysis and Wiener Filter of Barcode Subsections

While current systems already provide reliable results in correctly reading barcodes at close range, we want to increase the limits of these systems so they will be able to read these codes with extreme levels of blur and noise. Building off of the work by Christine Lew and Dheyani Malde[2], we propose a method to analyze a subsection of a barcode with an unknown signal and blurring function in order to learn their values. This function uses the Wiener filter to analyze and filter out noise and blur from a blurry barcode, while using a dictionary brute force method to try every possible combination of the first two digits of a barcode to find a blurring function and clean signal that created the original input. Through testing, our method successfully analyzes a subsection of a blurry barcode and returns a fragment of a clean barcode along with a kernel estimate similar to the actual function that blurred the image, possibly allowing future processes to accept even more limited data and correctly reconstruct the original data.