V.A. Boykov - Post-graduate Student, "Laser and Optoelectronic Systems" Department, Bauman Moscow State Technical University. E-mail: bazzilio215@rambler.ru V.Ya. Kolyuchkin - Dr.Sc. (Eng.), Professor, "Laser and Optoelectronic Systems" Department, Bauman Moscow State Technical University. E-mail: vkoluch@bmstu.ru

Development of reliable algorithms for object image selection, which are invariant to object image, background and noise parameters, and furthermore can provide a real-time image processing, is a relevant problem. Random Ferns method seems suitable for development of algorithms with the similar qualities. In this paper we propose an object image selection algorithm, based on random ferns. Algorithm requires initial training. We carried out the efficiency of such algorithm for several types of objects. Object images were exposed on spatially- regular or irregular background. We blur images with white Gaussian noise in purpose to vary signal-to-noise ratio. Four efficiency indexes of the algorithm were estimated for the each object image on several signal-to-noise ratio values: correct and dummy selection probabilities, and average computation time in case of bounding the processing area in image, and without bounding. We show that the object image selection algorithm, based on random ferns, in case of signal-to-noise ratio is greater than 15, provides reliable object image selection even on spatially-irregular background, i.e. correct selection probability is near to 100%, and dummy selection probability is equal to zero. Algorithm provides high performance. In case of signal-to-noise ratio is greater than 10, average computation time didn-t exceed 20 ms even for processing a whole image, which dimensions are equal to 1024*1024, and which contains spatially-irregular background and similar object images. Consequently, it is suitable for real-time working computer vision systems.

 

1. Alpatov B.A., Babajan P.V., Balashov O.E., Stepashkin A.I. Metody avtomaticheskogo obnaruzhenija i soprovozhdenija obektov. Obrabotka izobrazhenijj i upravlenie. M.: Radiotekhnika. 2008. 176 c.
2. Alpatov B.A., Babajan P.V. Metody obrabotki i analiza izobrazhenijj v bortovykh sistemakh obnaruzhenija i soprovozhdenija obektov // Cifrovaja obrabotka signalov.2006.№ 2.S. 45-51.
3. Babajan P.V. Vydelenie obektov v uslovijakh sluchajjnykh geometricheskikh preobrazovanijj izobrazhenija // Avtometrija.2010.№ 3.S. 41-47.
4. Primenenie metodov segmentacii izobrazhenijj v avtonomnykh sistemakh obnaruzhenija, raspoznavanija i soprovozhdenija dvizhushhikhsja celejj / Pod red. P.A. Bakuta // Zarubezhnaja radioehlektronika. 1987. № 10. S. 3-93.
5. Alpatov B.A., Blokhin A.N., Muravev V.S. Algoritm obrabotki izobrazhenijj dlja sistem avtomaticheskogo soprovozhdenija vozdushnykh obektov // Cifrovaja obrabotka signalov. 2010.№ 4.S. 12-17.
6. Muravev V.S., Muravev S.I. Adaptivnyjj algoritm vydelenija i obnaruzhenija vozdushnykh obektov na izobrazhenii // Informacionno-upravljajushhie sistemy. 2011.№ 5. S. 8-14.
7. Kalal Z., Mikolajczyk K., Matas J. Tracking-Learning-Detection // IEEE Transactions on pattern analysis and machine intelligence. 2012. V. 34. № 7. P. 1409-1422.
8. Viola P., Jones M. J.Rapid object detection using a boosted cascade of simple features // IEEE Conf. on Computer Vision and Pattern Recognition. Kauai, Hawaii, USA, 2001. V.1. P. 511-518.
9. Ozuysal M., Calonder M., Lepetit V., Fua P. Fast Keypoint Recognition using Random Ferns // IEEE Transactions on pattern analysis and machine intelligence. 2010. V. 32. № 3. P. 448-461.
10. Calonder M., Lepetit V., Fua P. BRIEF: Binary Robust Independent Elementary Features // Computer Vision - ECCV-10. 2010. R. 778-792.
11. Mastov A., Konovalenko I.A., Grigorev A.S. Adaptivnyjj podkhod k raspoznavaniju obektov s proizvolnogo rakursa v realnom vremeni // Trudy 39-jj shkoly-konferencii ITiS 2015. [EHlektronnyjj resurs] URL: http://itas2015.iitp.ru/pdf/ 1570177051.pdf (data obrashhenija 07.12.2015).
12. Krasilnikov N.N. Cifrovaja obrabotka 2D- i 3D- izobrazhenijj: ucheb. posobie. SPb.: BKHV-Peterburg. 2011. 608 c.