An inverse problem arising in decoding of bar codes

Series: 
School of Mathematics Colloquium
Thursday, January 20, 2011 - 11:00
1 hour (actually 50 minutes)
Location: 
Skiles 005
,  
University of Minnesota (Minneapolis)
Information encoded in a bar code can be read using a laser scanner or a camera-based scanner. For one-dimensional bar codes, which are in most prevalent use, the information that needs to be extracted are the widths of the black and white bars. The collection of black and white bars may be viewed as a binary one-dimensional image. The signal measured at the scanner amounts to the convolution of the binary image with a smoothing kernel. The challenge is that the smoothing kernel, in addition to the binary image, is also unknown. This presentation will review the technology behind bar code scanning and present several approaches to the decoding problem.