An Improvement on QR Code Limit Angle Detection using Convolution Neural Network

The QR Code implementation has been used in various fields such as manufacturing, mining, and logistics, retail, health, life sciences, transportation, and office automation. The advantage is useful for sharing information and integrated with a browser to get easier detail information. QR Code detection requires the position of the distance and the right reading angle. Moreover, for robots and the visually impaired person will not be able to do easily without guidance. The low-resolution camera cannot detect QR Code accuracy and need closer distance to be able to scan the QR Code. This is the challenge for improving the performance QR Code detection on low-resolution camera. This work proposes an algorithm that increases the reading of the angle of QR codes and maximizes detection on low-resolution cameras. This detection involves the image processing and Convolutional Neural Network (CNN) algorithm. This algorithm is applied to recognize the different QR Code version. The proposed approach can detect QR Code from a wider Euler Angle. The results demonstrated that the proposed detection method can increase 40o angle detection from the low- resolution and 15o angle detection from the higher-resolution. The proposed approach reached the accuracy rate of over 90%. These results revealed that the proposed method has high potential to to detect QR Code from the wider perspective angle.