Abstract: When QR code images are filmed by the camera, the deviation of shooting angle will cause the collected QR code images generate geometric distortion and bring difficulties to QR code recognition. Local threshold method and mathematical morphology are used to binarize the collected QR code images which appears uneven illumination. Harris corner detection algorithm and convex hull algorithm are combined to find the outline of the QR code and the point on the outline, using corner detection algorithm of this paper to find the corner of QR code. Finally, using perspective transformation algorithm to recover the distorted image, the results show that this method is effective to solve the problem of geometric distortion of the QR code.

Key words: binarization, corner detection algorithm, convex hull algorithm, perspective transformation algorithm

摘要： 用摄像机拍摄QR码图像时，由于拍摄角度的偏差，可能造成所采集到的QR码图像产生几何失真的情况，给QR码的识别带来了困难。对拍摄到的QR码图像进行预处理，把采集到的光照不均匀的QR码图像采用局部阈值法和数学形态学的方法进行二值化。通过Harris角点检测算法和凸包算法相结合找到QR码的轮廓以及轮廓上的点，再利用该角点检测算法找到QR码的角点，最后使用透视变化法对畸变图像进行恢复。实验结果证明，该方法有效解决了QR码的几何失真问题。

关键词: 二值化, 角点检测算法, 凸包算法, 透视变换算法