Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2016, Vol. 13, No. 5, pp. 304-311
Method of edge detection of objects in satellite images using minimal computing resources
E.P. Petrov
1 , N.L. Kharina
1 , K.N. Chukaev
1 1 Vyatka State University, Kirov, Russia
Accepted: 07.07.2016
DOI: 10.21046/2070-7401-2016-13-5-304-311
Edge detection in digital images is the initial and decisive phase in object identification in computer vision systems, Earth remote sensing and other systems. Most of the edge detection algorithms are based on calculation of the gradient image. The most common are Roberts’ and Prewitt operators. The advantage of these methods is low computational complexity. The disadvantages are high noise-sensitivity, possibility of discontinuity in the circuit. CANNY edge detection algorithm makes it possible to correctly determine the position of the boundary and gives only one response to the border, but it is difficult to implement. The choice of method depends on the ultimate goal of the entire image processing, types of processed images, available computing power. In contrast to the considered methods, the proposed method is based on a mathematical model of the image based on two-dimensional Markov chain, and it requires minimal computing resources for its realization. Moreover, it does not require the use of full-size images; two senior bit binary images combined into a plane are enough. Analysis of the results shows that the most close to the benchmark are obtained by CANNY method and the proposed method. However, CANNY method requires considerable computing resources for its realization making difficult its use in real-time image processing systems.
Keywords: digital image, satellite image, Markov random field, Markov chain, matrix of probabilities of one-step transition, element of image prediction, segmentation, object contour
Full textReferences:
- Ivanov E.S., Nekotorye prilozheniya segmentatsii snimkov DZZ (Some applications of segmentation of pictures of DZZ), Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2016, Vol. 13, No. 1, pp.105–116.
- Petrov E.P., Kharina N.L., Rzhanikova E.D., Matematicheskaya model' tsifrovykh polutonovykh izobrazhenii na osnove tsepei Markova s neskol'kimi sostoyaniyami (Mathematical model of digital grayscale images on the basis of Markov's chains with several states), Nelineinyi mir, 2013, Vol. 11, No. 7, pp. 487–492.
- Petrov E.P., Kharina N.L., Medvedeva E.V., Matematicheskaya model' tsifrovykh polutonovykh izobrazhenii Zemli iz kosmosa (Mathematical model of digital half-tone pictures of the earth from space), Proc. 2nd All-Russia Scientific and Technical Conference “Aktual'nye problemy raketno-kosmicheskoi tekhniki” (II Kozlovskie chteniya), Samara, 2011, pp. 179–185.
- Fisenko V.T., Fisenko T.Yu., Komp'yuternaya obrabotka i raspoznavanie izobrazhenii (Computer processing and recognition of images), Saint Petersburg: SPbGU ITMO, 2008, 192 p.
- Kharina N.L, Metod vydeleniya konturov na sputnikovykh snimkakh (Method of allocation of contours in satellite pictures), Proc. 2nd All-Russia Scientific and Technical Conference “Aktual'nye problemy raketno-kosmicheskoi tekhniki (IV Kozlovskie chteniya), Samara, 2015, pp. 258–262.