Development of an image segmentation method from unmanned aerial vehicles based on the ant colony algorithm under the influence of speckle noise

The object of research is the process of segmenting an image from an unmanned aerial vehicle based on the ant algorithm under the influence of speckle noise. Unlike the known ones, the image segmentation method based on the ant algorithm involves the imitation of the collective behaviour of agents (ants) capable of adapting to local features of the image. In addition, the pheromone marking mechanism contributes to a more distinct delineation of the boundaries between segments, which positively affects the accuracy of dividing the image into segments. Speckle noise is a type of multiplicative noise that occurs in images formed using coherent radiation. Its appearance is due to the interference of reflected waves coming from different points of the same object, but with microscopic differences in phase. This leads to the appearance of a chaotic granular structure that distorts the image and complicates further analysis. Experimental studies have shown that the segmentation method based on the ant algorithm provides a reduction in segmentation errors of the first kind on average from 6% (in the absence of speckle noise) to 30% (at a speckle noise intensity σ = 15). With an increase in the speckle noise intensity, the gain in the value of the segmentation error of the first kind increases. The segmentation method based on the ant algorithm provides a reduction in segmentation errors of the second kind on average from 5% (in the absence of speckle noise) to 32% (at a speckle noise intensity σ = 15). With an increase in the speckle noise intensity, the gain in the value of the segmentation error of the second kind increases. The practical value of the segmentation method based on the ant algorithm lies in the possibility of segmentation under the influence of speckle noise. At the same time, a reduction in segmentation errors of the first and second kind is ensured in comparison with the known method.