Improving image stitching methods for mapping low texture wind turbine blades

22 de junho de 2022

The world's ever-growing energy needs have directed focus on the deployment of alternative renewable sources, namely wind energy. The blade is one of the most important components for the Wind Turbine to capture energy effectively. Therefore, they need to be continuously monitored. Researchers have introduced a safe, automated, and reliable alternative to inspect the blade: visual inspection through Unmanned Aerial Vehicles, more commonly called drones. In order to detect small damages on the blade, the images acquired need to have high resolution. This Thesis focuses on developing an end-to-end system that automatically produces a high-resolution mosaic of a Wind Turbine blade. The system takes as input the multitude of images acquired by the drone during the inspection and merges them into a single one. However, two major challenges had to be overcome: the distracting background of wind farms and the featureless surface of a blade. To overcome these problems, it is proposed the GraphCut algorithm and Contrast Limited Adaptive Histogram Equalization method for Image Enhancement, respectively. The final mosaic obtained by the tailored made solution looks seamless in geometry, i.e. the edges of the blade are aligned and there are no discontinuities. A commercially available software called AutoStitch was not able to produce the mosaic.

Autores da comunidade :

• 

  I

  Inês Gaspar Coelho

  ist187022

Orientadores desta instituição:

• 

  M

  Manuel Fernando Cabido Peres Lopes

  ist40208

• 

  A

  Alexandre José Malheiro Bernardino

  ist13761

electrical-engineering-electronic-engineering-information-engineering - Engenharia Eletrotécnica, Eletrónica e Informática

eng - Inglês

Instituto Superior Técnico