Abstract:
In the year of the global pandemic there was an increment of the usage of QR code due to the development of the EU Digital COVID Certificate, among others type of certificate of other countries. This lead us to the problem of mobile phone having difficulties to reading the QR code. In this thesis, we evaluate the QR code detection and decoding performance of a popular open-source library by applying different image noise models. Our approach works by simulating several image degradation factors like thermal noise, perspective distortion, defocus, and Moirè patterns originated when capturing an LCD screen. Experimental results show that the detection part plays a significant role and, surprisingly, the error-correction capability of the marker might be inversely proportional to the decoding rate.
