Key points are not available for this paper at this time.
Abstract Meta-Heuristic (MH) algorithms have gained prominence in computer vision and image processing due to their efficacy in solving complicated, high-dimensional optimization challenges. This review study thoroughly evaluates the effectiveness of MH approaches in classification, segmentation, and registration applications. The compilation consists of 84 studies: 39 in classification (47%), 23 in segmentation (27%), and 22 in registration (26%). The examination of these investigations reveals that the implementation of MH algorithms in hybrid models utilizing deep learning offers notable benefits in enhancing accuracy, circumventing local optima, and decreasing computational expenses. This research also examines limitations, including the substantial computing demands in real-time applications and the challenges related to data processing. The paper highlights the significant potential of MH algorithms in healthcare, agriculture, security, and remote sensing, along with their role in addressing current challenges. Renowned international publishers, such as Elsevier, Springer, IEEE, and MDPI, have disseminated relevant contemporary research. The acceptance percentages for these publications are 42%, 24%, 12%, and 11%, respectively. Publications from alternative publishers account for the remaining 11%. Also, the source codes and associated datasets of the 84 studies examined in this paper are available as open source at this link: https://github.com/mfaruk-sahin/Metaheuristics-in-Image-Processing-and-Computer-Vision.git
Sahin et al. (Mon,) studied this question.