Barcode Detection and Decoding in On-Line Fashion Image
Abstract
A barcode is the representation of data including some information related to goods, offered for sale which frequently appears in markets. Especially in the online fashion market such as the buy and sell market, barcodes on the tags of the sale items support identified information including producer, manufacturer, etc. The market need a system to automatically detect and decode barcode in real time. However, the existing method has a limitation in detecting 1-D barcode in some backgrounds such as tassels, stripes, and texture in fashion images. In this research, a focus is on identifying the barcode and distinguishing a barcode from its similarities. It is accomplished by adding a post-processing technique after morphological operations in the traditional method based on the hand-crafted features. Convolution Neural Network (CNN) is applied to solve this typical objective detection problem. The proposed algorithm has been validated using several examples. In addition, the performance and the results of the proposed algorithm have been compared with the other methods presented in the literature. To decode a barcode, a Python-supported package including the existing common types of decoding schemes is widely used to decode the barcode. However, this commonly-used package has limitations in decoding the skewed barcodes. A preprocessing transformation step is added to process the strongly skewed barcode images in order to improve the probability of decoding success.
Degree
M.Sc.
Advisors
Allebach, Purdue University.
Subject Area
Artificial intelligence|Commerce-Business|Pedagogy
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.