A General Scheme for Extracting QR Code from a non-uniform background in
Camera Phones and Applications
Yu-Hsuan Chang, Chung-Hua Chu and Ming-Syan Chen
Department of Communication Engineering
National Taiwan University
Taipei, Taiwan, ROC
E-mail: sharon, jony@arbor.ee.ntu.edu.tw, mschen@cc.ee.ntu.edu.tw
Abstract
With the rapid advances in mobile communication tech-
nologies, QR code in the embedded camera devices has
been used as new input interfaces. However, the pr evious
works for extracting Q R code from an image do not consider
a non-uniform background. In this paper, we implement the
applications of QR code and propose an efficient algorithm
to extract QR code from the non-uniform background. In
contrast with prior works, our approach is of higher accu-
ra cy for QR-code recognition and more practical for use in
a mobile information en vironment.
1 Introduction
QR code has been used in various applications since the
QR code has the large capacity, the small printout size, the
high speed scan, the damage resistance and the data robust-
ness [5]. There are many advantages to use the QR code in
mobile phones such as omni-direction readability and error
correction capability. Therefore, mobile phones adopt the
QR code to support many services such as booking tickets,
paying a fee and URL reading [8] [12] [14].
The conventional approaches [2] and [7] are to extract a
barcode from the image of a uniform and even background.
However, in the real circumstance, the light condition of
the image taken from the camera phone is uneven, and the
background of the image is non-uniform. Therefore, the
traditional schemes suffer from accurate issues in the real
environment.
The problem we consider is referred to as the Extracting
QR code Problem. Given an image with uneven light in
a non-uniform background taken from camera phones, the
problem is to remove the bad light condition and extract
the QR code to improve the correctness of the recognition.
To the best of our knowledge, there is no previous work on
extracting the QR code from the images with non-uniform
background and uneven light taken from camera phones.
In this paper, we first propose the QR-code extraction al-
gorithm (QRE) to solve the problem. QRE first uses light
balancing to eliminate uneven light noise in the image. Af-
terward, QRE uses the features of the QR code to extract
the barcodes precisely. We also conduct several experi-
ments, and results show that QRE leads to the higher ac-
curacy as compared to other conventional schemes. Conse-
quently, extracting the barcodes from the images with non-
uniform background and uneven light is indeed a necessary
technique in a mobile information environment.
This paper is organized as follows. We describe the sys-
tem architecture and the proposed algorithm for the QR-
code extraction in Section 2. In Section 3, the experimental
results are shown. Finally, this paper concludes in Section
4.
2 QR-code Extraction Algorithm
2.1 QR-code Structure
QR code is a matrix 2D barcode. Each module corre-
sponds to a bit and a black module and a white module rep-
resent ‘1’and‘0’ respectively. Figure 1 (a) illustrates the
structure of the QR code. There are three finder patterns lo-
cated at the corners. Since these finder patterns can be used
for position detection, the QR code is capable of 360 de-
gree (i.e., omni-directional) and high speed reading. Figure
1 (b) shows the area allocation of the QR code. Mainly, the
square area is divided into three regions as the error correc-
tion area, the data area and the reserved area.
2.2 System Architecture
Figure 2 depicts our system architecture. Our architec-
ture contains a web-based information system and an im-
Ninth IEEE International Symposium on Multimedia 2007
0-7695-3058-3/07 $25.00 © 2007 IEEE
DOI 10.1109/ISM.2007.26
123