Enhanced Image Differencing for Precise Change Detection in Static Environments
Yohei Nishidate, Yukihide Kohira, Shigeo Takahashi, Rentaro Yoshioka · Information Engineering Express · 2024
This study addresses the challenge of detecting subtle changes in static environments, specifically focusing on images taken at different times in a museum.Typically, the method based on direct image differencing and thresholding for change detection encounters limitations due to high rates of false positives and difficulty in discerning subtle c hanges.To improve accuracy, we employed enhanced correlation coefficient (ECC) maximization with a perspective transformations model for image alignment.Also, we developed a color adjustment methodology combining Lab color scale conversion with CLAHE equalization in harmonizing color intensity under varied lighting conditions.Experiments conducted at the Fukushima Prefectural Museum with images from 11 locations demonstrated the effectiveness of our methods in detecting a range of changes, from object displacements to lighting variations.The study highlights the potential of these techniques in applications requiring precise change detection in static settings, with recommendations for future work aimed at refining these approaches for broader scenarios and challenging lighting conditions.