Segmentation of optical-electronic images from on-board systems of remote sensing of the earth by the artificial bee colony method
Ігор Вікторович Рубан, Hennadii Khudov, Олександр Миколайович МАКОВЕЙЧУК, Irina Khizhnyak, Vladyslav Khudov, Viacheslav Podlipaiev, Viktor O. Shumeiko, Oleksandr Atrasevych, Anatolii Nikitin, Rostyslav Khudov · Eastern-European Journal of Enterprise Technologies · 2019
It was established that it is not possible to apply the known methods of image segmentation directly to segmentation of optical-electronic images of on-board systems of remote sensing of the Earth. We have stated the mathematical problem on segmentation of such images. It was established that the result of segmentation of images of on-board systems of remote sensing of the Earth is separation of an image into artificial objects (objects of interest) and natural objects (a background). It has been proposed to use the artificial bee colony method for segmentation of images. We described the essence of the method, which provides for determination of agents positions, their migration, conditions for stopping of an iteration process by the criterion of a minimum of a fitness function and determination of the optimal value of a threshold level. The fitness function was introduced, which has the physical meaning of a sum of variance brightness of segments of a segmented image. We formulated the optimization problem of image segmentation of an on-board optical-electronic observation system. It consists in minimization of a fitness function under certain assumptions and constraints.The paper presents results from an experimental study on application of the artificial bee colony method to segmentation of an optical-electronic image. Experimental studies on segmentation of an optical-electronic image confirmed the efficiency of the artificial bee colony method. We identified possible objects of interest on the segmented image, such as tanks with oil or fuel for aircraft, airplanes, airfield facilities, etc.The visual assessment of the quality of segmentation was performed. We calculated errors of the first type and the second type. It was established that application of the artificial bee colony method would improve the quality of processing of optical-electronic images. We observed a decrease of segmentation errors of the first type and the second type by the magnitude from 7 % to 33 % on average