Developing the algorithm of the machine vision system for the images semantic analysis

Authors: Vasileva E.A.
Published in issue: #9(26)/2018
DOI: 10.18698/2541-8009-2018-9-366
Category: Mechanical Engineering and Machine Science | Chapter: Robots, Mechatronics, and Robotic Systems
Keywords: machine vision, unmanned aerial vehicle, group control, pattern recognition, machine learning, pattern classification, neural network, algorithm SVM
Published: 04.09.2018

The article investigates the problem of controlling the agricultural sector autonomously and suggests using a group of robots in order to solve this problem. We determine an appropriate strategy of group control. The authors consider the task of detecting the oppressed parts of fields with the help of unmanned aerial vehicles. In order to recognize the images and reduce the objects to semantic models we select a combination of methods based on the pretrained neural network VGG16 developed by a group of scientists from Oxford and the machine-learning algorithm SVM (Support Vector Machine). We have developed a pattern recognition and classification algorithm as well as the software implementing this algorithm. The suggested algorithm is written in the computer programming language Python. The article describes the pattern recognition algorithm and introduces the results of testing the operational efficiency of the developed system. It is established that the development of the external environment’s semantic models may significantly increase the efficiency of solving the navigation problems when controlling a group of robots autonomously.

References

[1] Kalyaev I.A., Rubtsov I.V. Combat robots need a program. Natsional‘naya oborona, 2012, no. 8(77), pp. 34–48.

[2] Sheremet I.B., Rudianov N.A., Ryabov A.V., Khrushchev V.S., Komchenkov V.I. Justification of combat and support a family of robots to fight in the city. Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2012, no. 3(128), pp. 31–41.

[3] Kalyaev I.A., Gayduk A.R., Kapustyan S.G. Modeli i algoritmy kollektivnogo upravleniya v gruppakh robotov [Models and algorithms of collective control in robot groups]. Moscow, Fizmatlit publ., 2009, 278 p.

[4] Rubtsov I.V., Mashkov K.Yu., Naumov V.N. 185 let MGTU im. N.E. Baumana – Sostoyanie i perspektivy razvitiya spetsial‘noy robototekhniki [185th anniversary of Bauman MSTU – present state and prospects of special robotics development]. Sb. mat. 10y Vseross. nauch.-prakt. konf. Perspektivnye sistemy i zadachi upravleniya [Proc. 10th Russ. Sci.-Pract. Conf. “Perspective control systems and problems”], 2015, no. 1, pp. 247–255.

[5] Dobbs T. Farms of the future will run on robots and drones. Available at: http://www.pbs.org/wgbh/nova/next/tech/farming-with-robotics-automation-and-sensors/ (accessed 27 May 2018).

[6] Mestetskiy L.M. Matematicheskie metody raspoznavaniya obrazov [Mathematical methods of image recognition]. Available at: http://www.cs.ru/frc/papers/mestetskii04course.pdf (accessed 27 May2018).

[7] Tarkhov D.A. Neyrosetevye modeli i algoritmy [Neural network models and algorithms]. Moscow, Radiotekhnika publ., 2014, 352 p.

[8] Chen L.-C., Papandreou G., Kokkinos I., Murphy K., Yuille A.L. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, vol. 40, no. 4, pp. 834–848.

[9] Cristianini N., Shawe-Taylor J. An introduction to support vector machines and other kernel-based learning methods. Cambridge University Press, 2000, 204 p.

[10] Vapnik V.N. Vosstanovlenie zavisimostey po empiricheskim dannym [Relations restoration using empirical data]. Moscow, Nauka publ., 1979, 448 p.