Remote QR Code Recognition Application: Solar Panels Identification in Solar Plant

The indexation of solar panels in solar plants is inescapable to ensure an efficient monitoring of solar panels. In this work, we propose a novel system which is able to remotely sense multiple Quick Response codes (QR) used for panels indexing. After acquiring an HD image of a scene which may contain QR codes, the system converts the image to a standard definition (SD) in order to decrease the computation time. Then, enhancing its contrast using the Look-up table transform. Once established, the resulting image is converted to the binary representation. Therefore, the system looks for the finder patterns (FPs) of the QR codes. FPs are three squares placed in the three corners of each QR code and respecting a specific ratio 1:1:3:1:1. These patterns are localized after browsing the image vertically and horizontally, so as to retain only the segments which respect the evoked ratio. The intersection of the two brows yields number of candidate areas (may contain FPs). These areas are then passed down to the next stage which aims removing the isolated pixels which cause false positives. Once achieved, the system regroups the remaining areas in such a way that each group will contain nearby areas according to a predefined distance, hence, each group assumed to form a QR code. Afterwards, a perspective correction is applied for the deformed QR codes. Finally, a QR code decoding algorithm is called so as to extract the embedded data of each QR code. The proposed system is able to localize multiple QR codes from a distance of more than 3 meters through images having a complex background (texture). Unlike the conventional system, which performs it for no more than 20 cm and in homogenous background. Thereby, the proposed system is adapted for remote solar panels identification, and the obtained results are encouraging.