Image Pre-processing and Quality Analysis Using Structural Similarity Indexing for Drone Applications

Abstract

The increasing popularity of drones in mapping, surveying, and inspection processes underscores their potential in various industrial applications. Despite their widespread use, the accuracy and reliability of drone imagery for solar panel inspection have not been extensively verified. This paper introduces a novel image quality assessment method for drone applications, utilizing the Structural Similarity Indexing Method (SSIM). The proposed methodology evaluates the quality of images captured by drones, focusing on the drone's positional relationship to the solar panels under diverse conditions. The study compared the quality of images with those subjected to Gaussian and Median filtering. The SSIM was employed as the primary metric to quantify the similarity between the original and processed images, providing a robust measure of image quality degradation or enhancement. In indoor tests, the SSIM values consistently decrease as the height increases from 0.7m to 2.5m for both Median and Gaussian filters. Meanwhile, outdoor tests reveal that the image achieves its highest SSIM score at a height of 1.0m for both Median and Gaussian filters. It is important to note that the Gaussian filter consistently yields slightly higher SSIM values compared to the Median filter at all heights. The SSIM analysis revealed significant insights into the optimal conditions for drone imaging in solar panel inspections. The findings of this research contribute to the development of standardized practices for drone-based inspections, ensuring high accuracy and reliability.