Development of an image segmentation method from unmanned aerial vehicles based on the particle swarm optimization algorithm

The object of research is the process of segmenting images from an unmanned aerial vehicle based on the particle swarm algorithm. One of the most problematic areas in segmenting images from unmanned aerial vehicles is the reduction in the efficiency of known segmentation methods. In addition, most methods do not accurately recognize small objects that occupy a small part of the image. The method of segmenting images from an unmanned aerial vehicle based on the particle swarm algorithm has been improved, in which, unlike the known ones, the following is performed: – the source image is converted to the appropriate color space; – the channel is selected for further analysis; – the particle swarm is initialized on the source image in each channel selected for further analysis; – the objective function is calculated for each particle of the swarm in the image in each selected channel; – the current value of the objective function for each particle of the swarm is compared with the best value of the objective function in the image in each selected channel; – calculating the velocity value and new location for each swarm particle in the image; – moving each swarm particle in the image in each selected channel; – determining the swarm particles with the best value of the objective function in the image in each channel; – combining the channels and forming the resulting image. During the study, it was found that the segmented image by the improved method based on the particle swarm algorithm has better visual quality compared to the known segmentation method. It was found that the improved segmentation method based on the particle swarm algorithm provides an average reduction in segmentation errors of the I kind by 11% and an average reduction in segmentation errors of the II kind by 9%.