Structure analysis through multispectral image processing from a UAV

For the industry, the inspection of structures has become a tedious task which requires the investment of an important amount of time and resources, yet necessary to guarantee the proper safety measures demanded by laws. With each new improvement in the structural analysis, companies keep trying to find faster and more reliable ways to check their buildings without risking their employees or having to rely on expensive methods of inspection for accurate results. This graduate thesis focuses on the development of a platform for aerial inspection based on multispectral image processing. An unmanned aerial vehicle performs both the navigation and vision processes through an embedded microcomputer and a multispectral sensor that captures images of the structure. The drone is positioned over the target through semiautonomous commands before starting the analysis. Different spectral images of the structure are collected to extract features and use them along with the spectral signatures of the possible failures to classify each flaw on the inspected section. The results of the failure detection are sent to a ground control station where the operator is able to find should there be problems on the structure, hence reducing the time required for the inspection and the amount of resources and workforce required.