CONSTRUCTION OF METHODS FOR DETERMINING THE CONTOURS OF OBJECTS ON TONAL AEROSPACE IMAGES BASED ON THE ANT ALGORITHMS

A method has been proposed for determining contours of objects on tonal aerospace images based on ant algorithms. The method, in contrast to those already known, takes into consideration patterns in the image formation; the ant algorithm is used for determining the contours. Determining an object's contours in the image has been reduced to calculating the fitness function, the totality of agents' motion areas, and the pheromone concentration along agents' motion routes. We have processed a tonal image for determining the contours of objects using a method based on the ant algorithm. In order to reduce the number of "junk" objects, the main principles and stages of the method for multi-scale processing of aerospace images based on the ant algorithm have been outlined. Determining the contours on images with a different value of the scale factor is carried out applying a method based on the ant algorithm. In addition, we rescale images with a different scale factor value to the original size and calculate the image filter. The resulting image is a pixelwise product of the original image and the image filter. The multiscale processing of tonal aerospace images with different scale values has been performed using methods based on the ant algorithms. It was established that application of a multi-scale processing reduces the number of "junk" objects. At the same time, due to multi-scale processing, not the objects' contours are determined but the objects in full. We estimated errors of first and second kind in determining the contours of objects on tonal aerospace images based on the ant algorithms. It was established that using the constructed methods has made it possible to reduce the first and second kind errors in determining the contours on tonal aerospace images by the magnitude of 18–22 % on averag