Stitching Approach for PV Panel Detection

Solar photovoltaic installations are exponentially increasing due to their cost competitiveness and low emissions. Regular inspection of PV panels ensure they deliver the expected power and reduce potentially long-term damages. PV panels can be inspected manually, semi-autonomously, and fully autonomously. This paper utilizes an Unmanned Aerial Vehicle (UAV) to fly over the PV panels and complete inspections fully autonomously. The UAV collects thermal and regular video, that is post-processed through the proposed novel algorithm. The proposed algorithm converts the video into images and groups the images in sets. Each set is stitched, and then the group of stitched images is stitched one final time to produce one complete high-resolution image. A series of additional image processing steps identify the PV panels and label them individually.