An algorithm dedicated to the detection and decoding of an alphabetic 2D barcode, namely the Alphacode, from a captured image is developed. When the camera is freely handled, images are degraded by several artifacts motion, acquisition, low resolution...), making recovery of the code a more difficult task. We propose a two-stage strategy. The code structure is first recovered using image processing steps including localization, geometric correction and code grid structure retrieval. The code is then recognized using an iterative structure based on a factor graph representation of the displaying/reading process and making use of the specific structure of the code. The proposed algorithm is evaluated in case of strongly blurred images and shows high improvements in performances compared to the basic processing which is unable to decode reliably.