Parallel optimization algorithms in combining swarm uav images

The use of Unmanned Aerial Vehicles (UAVs) in the defense industry and civilian applications, particularly with the development of swarm UAV systems, has increased the need for image mosaicing. In this study, Mean Square Error (MSE) and Normalized Cross Correlation (NCC) were used as objective functions to measure the level of similarity in overlapping regions of images. The main contribution of the thesis addresses the problem of simultaneous fusion of multiple images. This problem exhibits the multi-objective characteristic of requiring a separate and independent search space for each image pair. To overcome this challenge, the Multi-Hive Artificial Bee Colony (MH-ABC) and Multi-Cavity Artificial Locust Swarm (MT-ALSO) algorithms, which are multi-swarm versions of classical algorithms, were proposed and developed. The proposed multi-swarm algorithms were applied to a complex problem consisting of nine real sunflower field images captured by a drone. The findings prove that MH-ABC and MT-ALSO algorithms are applicable and effective in solving such complex image fusion problems with multiple and parallel search spaces, and successfully combine the fragmented images to form an integrated mosaic.