This thesis describes a novel method for decoding linear barcodes from blurry camera images. lt can be used on mobile devices to recognize product numbers from EAN or UPC barcodes and retrieve information from the Internet without input from the user. Conventional decoders for linear barcodes are based on detecting the edges between bars and spaces. However, the locations of these edges may change or become undetectable if the input is very blurry. This is a problem for mobile devices with fixed-focus lenses. Their cameras are not designed to focus correctly in the macro range required for capturing barcodes. The Apple iPhone is a well-known example of such a device. The proposed algorithm locates the barcode in the camera image and extracts a scan line of brightness values. lt simulates the blurry barcode according to a mathematical model and chooses digits for which the simulation best approximates the camera input. The parameters of the simulation are adjusted automatically with an iterative approach. A prototype was implemented to recognize UPC-A and EAN-13 barcodes on the Apple iPhone and on the MacBook, using their built-in cameras. The decoder was tested with several hundred images from different cameras. The proposed method correctly recognizes a high percentage of blurry barcode images. The performance of the prototype is also compared to four different existing decoders for linear barcodes, with good results.