Deblurring of Images and Barcode Extraction of PV Modules using Supervised Machine learning for Plant Operation and Maintenance

This paper describes a supervised machine learning based automated barcode detection technique for fast re-creation of spatial mapping of module serial numbers in a plant. Such information is crucial while performing the drone based visual and infrared inspections of the plant and it can result in significantly reduced time and more effective operation and maintenance. The performance of image processing-based technique and a supervised machine learning based technique is compared in this paper. The described technique is robust and generic and therefore, can be used to automatically detect other features in the module such as snail trails, discolored cells etc.