摘要
采集的QR码图像可能存在光照不均、旋转、扭曲和噪声等现象,位置探测图形难于满足标准深浅模型宽度比1∶1∶3∶1∶1。基于位置探测图形定位法是通过设置的容差来判定是否为位置探测图形,但对大信息量和低分辨率的QR码图像定位率较低。针对此问题,提出基于亚像素的QR码定位法。首先对位置探测图形设置较大容差进行粗定位,然后设置较小容差进行精定位;对精定位失败的,对其过渡像素分别使用最大最小值法、平均值法和中值法进行分解,再以同容差进行精定位。实验对比表明,提出的3种过渡像素分解方法都提高了定位率,以平均值法为最优,在旋转的分辨率为46~88像素的QR码图像集中,比基于位置探测图形定位法平均提高了24.5%。
Acquired QR code image may have the phenomena of uneven illumination,rotation,distortion and noise,so the finder patterns are difficult to meet standard shades model width ratio 1： 1： 3： 1： 1. The locating method based on finder patterns determines whether to be finder patterns by setting the tolerance. However,it is limited when applied to the information-heavy and low-resolution QR images for the low localisation ratio. This paper presents a sub-pixel-based QR code locating method for this problem. Firstly,it sets bigger tolerance on finder patterns for coarse locating,and then sets a smaller tolerance for precise locating. For failure in precisely locating the patterns,their transition pixels are to be decomposed using max-min value method,average value method and median method respectively,and then the same tolerance will be used to operate fine location once again. Contrast experiments show that the proposed three transition pixel decomposition methods all improve the localisation rate,but the average locating method is the best. In the QR code image set with 46 ~ 88 px rotation resolution,the method increases by 24. 5% in average over the locating method based on finder patterns.
出处
《计算机应用与软件》
CSCD
2015年第12期214-217,260,共5页
Computer Applications and Software
关键词
QR码
QR码定位
亚像素
像素分解
QR code
QR code localisation
Sub-pixel
Pixel decomposition
