Segmentation of the images obtained from onboard optoelectronic surveillance systems by the evolutionary method

Abstract

The essence of the simpler evolutionary method of image segmentation which relates to ant methods was set forth. The image segmentation process was presented as a set of areas in which agents (ants) move. Probability of transition from one turning point of the route to another was determined taking into account attractiveness of the route and concentration of pheromones on it. A timely convergence of decisions (choice of the same route by the agents) is processed by the use of feedback, i.e. evaporation of pheromones. The parameters setting pheromone weight and attractiveness of the area were calculated. The routes which are the most attractive according to the selected criteria (with the maximum concentration of pheromone) were determined. Unattractive routes disappear with a gradual drying of pheromone on such routes. When checking function ability of the simpler evolutionary segmentation method, it was found that implementations of this method with obviously unsuccessful results are possible. Essence of the advanced evolutionary method of image segmentation as improvement of the simpler evolutionary method was outlined. In the improved method, only the best agents increase the level of pheromone on their routes. The level of pheromone on the routes is limited. An expression has been obtained for renewal of pheromone levels. The best route may be either the iteration best or the best-so-far (found since the start of the method) route. In contrast to the simpler evolutionary method, an optimal route of agent movement was found during segmentation of images in all implementations with the use of the advanced evolutionary method. Experimental studies of segmentation of the images obtained from the onboard systems of optoelectronic surveillance using the evolutionary method have been carried out. As an example, possible objects of interest were defined in the segmented image and it was established that the outlined contours of the main objects of interest coincide with the boundaries of the objects in the original image. Presence of a large number of outlined contours of small-sized objects in the segmented image was pointed out and an example of such area was given. Visual estimation of efficiency of application of the evolutionary method was made