Blind Deblurring of QR Codes with Local Extremum Intensity Prior

In recent years, QR code has been widely applied in various industries and has provided great convenience for our daily life. However, due to some inevitable factors such as camera shake and defocus, the images captured by the camera sensor might become blurry, which makes the QR code difficult to be recognized and be decoded. To deal with this problem, we propose a blind image deblurring method based on a novel binary prior named Local Extremum Intensity Prior (LEI) for QR code images. The proposed prior is based on the observation that the LEI of clear QR code images is sparser than that of blurred ones. By introducing a simple yet effective threshold technique to compute the LEI-involved regularization term, together with an effective optimization scheme, our method can recover clean and readable results from the blurred QR code images. Extensive experimental results show that our proposed method has advantages in restoration quality compared with the state-of-the-art image deblurring methods.