A METHOD FOR DETECTION AND MATCHING OF IMAGE KEYPOINTS FOR OBJECT RECOGNITION AND TRACKING

  • V.A. Tupikov SPE "Research and Production Enterprise "Air and Marine Electronics"
  • V.A. Pavlova SPE "Research and Production Enterprise "Air and Marine Electronics"
  • V.A. Bondarenko SPE "Research and Production Enterprise "Air and Marine Electronics"
  • G.E. Kaplinskiy SPE "Research and Production Enterprise "Air and Marine Electronics"
Keywords: Image processing, key points, detector, descriptor, automatic object tracking

Abstract

The goal of this paper is to improve the efficiency of automatic object and landmark detec-tion and tracking algorithms in poor observation conditions in presence of noise and artificial obstacles. A particular complication for automatic tracking algorithms in these conditions are rapid scalar and affine image distortions of the object being tracked. In such situations the best performance is shown by the image processing algorithms based on image key features (key point) detection. To solve the stated task a new automatic object and landmark detection and tracking algorithm is developed, based on image key point detection techniques. The developed algorithm of several separated stages. The preliminary stage of processing of each image, being received from electro-optical system (EOS), is the preprocessing stage, which consist of image brightness and contrast equalization. On the first algorithm stage the key point detection and filtering is done with the use of highly efficient Hessian matrix approximation of multiscale second order discrete image derivatives. The local spatio-scale space maxima of Hessian matrix determinant are then selected as key points. Each found key point is then processed to build the description of its local neighborhood. This paper proposes a new multicomponent key point feature vector, providing fast preliminary key point comparison with more precise refinement on the later comparison stages if necessary. The last stage of algorithm consists of finding the best match between two key point sets of tracked object and current image being processed. In case when current video frame contains an object or its significant part, this algorithm allows to precisely locate its position and transfor-mation on the current image with the computational complexity sufficiently low to use the algo-rithm in embedded systems of manned and unmanned aerial vehicles. The proposed algorithm has been tested as a part of special purpose software onboard of state-of-the-art unmanned aerial vehicle and proved to be highly efficient and noise resistant.

References

1. Lowe D.G. Object recognition from local scale-invariant features // Proceedings of the Interna-tional Conference on Computer Vision. – 1999. – No. 2. – P. 1150-1157.
2. Sung-Joo Y., Kim T. Development of Stereo Visual Odometry Based on Photogrammetric Fea-ture Optimization // Remote Sensing. – 2019. – Vol. 11.
3. Grundmann M., Kwatra V., Castro D., Essa I. Calibration-free rolling shutter removal // In Proc. ICCP, 2012.
4. Brown M., Lowe D.G. Automatic Panoramic Image Stitching using Invariant Features // Inter-national Journal of Computer Vision. – 2007. – Vol. 1. – P. 59-73.
5. Attneave F. Some informational aspects of visual perception // Psychological Review. – 1954. – Vol. 61. – P. 183-193.
6. Rodehorst V., Koschan A. Comparison and evaluation of feature point detectors // Proceedings of 5th International Symposium Turkish-German Joint Geodetic Days, 2006.
7. Tuytelaars T., Mikolajczyk K. Local Invariant Feature Detectors – Survey. Foundations and Trends in Computer Graphics and Vision. – 2008. – No. 3 (3). – P. 177-280.
8. Jazayeri I., Fraser C. Interest Operators in Close-Range Object Reconstruction // The Interna-tional Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. – 2008. – Vol. XXXVII. – Part B5. – P. 69-74.
9. Přibyl B., Chalmers A., Zemčík P. Feature point detection under extreme lighting conditions // Proceedings of the 28th Spring Conference on Computer Graphics. – 2012. – P. 143-150.
10. Lindeberg T. Image matching using generalized scale-space interest points // Journal of math-ematical Imaging and Vision. – 2015. – Vol. 52, No. 1. – P. 3-36.
11. Zuiderveld K. Contrast limited adaptive histogram equalization // Graphics Gems IV. – 1994. – Chapter VIII.5. – P. 474-485.
12. Красильников Н.Н. Цифровая обработка 2D- и 3D-изображений: учеб. пособие. – СПб.: БХВ-Петербург, 2011. – 608 с.
13. Moravec H. Towards Automatic Visual Obstacle Avoidance // Proceedings of the 5th Interna-tional Joint Conference on Artificial Intelligence, Cambridge. 1977. – P. 584.
14. Harris C., Stephens M. A combined corner and edge detector // Proceedings of the 4th Alvey Vision Conference. – 1988. – P. 147-151.
15. Shi J., Tomasi C. Good features to track // Proceedings of IEEE Conference on Computer Vi-sion and Pattern Recognition. – 1994. – P. 593-600.
16. Förstner W. Gülch A. Fast Operator for Detection and Precise Location of Distinct Points, Corners and Centres of Circular Features // Proceedings of the ISPRS Intercommission Work-shop on Fast Processing of Photogrammetric Data. – 1987. – P. 281-305.
17. Smith S.M., Brady J.M. SUSAN–A New Approach to Low Level Image // Processing Interna-tional Journal of Computer Vision. – 1997. – Vol. 23 (1). – P. 45-78.
18. Bae S.C., Kweon I.S. COP: A new method for extracting edges and corners // Lecture Notes in Computer Science. – 1997. – P. 271-278.
19. Rosten E. Drummond T. Fusing points and lines for high performance tracking // IEEE Inter-national Conference on Computer Vision. – 2005. – Vol. 2. – P. 1508-1511.
20. Mikolajczyk K., Schmid C. A performance evaluation of local descriptors // PAMI. – 2005. – Vol. 27. – P. 1615-1630.
21. Gauglitz S., Höllerer T., Turk M. Evaluation of interest point detectors and feature descriptors for visual tracking // International Journal of Computer Vision. – 2011. – P. 1-26.
22. Макаров А.С., Болсуновская М.В. Сравнительный анализ методов обнаружения особых точек на изображениях при различных уровнях освещения // Научно-технические ведомо-сти СПбГПУ. Информатика. Телекоммуникации. Управление. – 2018. – № 11:2. – C. 7-18.
23. Mikolajczyk K., Schmid C. An affine invariant interest point detector / In: Heyden, A., Sparr, G., Nielsen, M., Johansen, P. (eds.) ECCV 2002. LNCS. – Vol. 2350. – P. 128-142.
24. Mikolajczyk K., Schmid C. Scale and Affine invariant interest point detectors // In IJCV. – 2004. – Vol. 60 (1). – P. 63-86.
25. Lowe D.G. Distinctive image features from scale-invariant keypoints // Int. J. Comput. – 2004. – Vis. 60 (2). – P. 91-110.
26. Bay H., Ess A., Tuytelaars T., Van Gool L. SURF: speeded up robust features // Comput Vis Image Understand. – 2008. – P. 346-359.
27. Crow F. Summed-area tables for texture mapping // In Proceedings of SIGGRAPH. – 1984. – Vol. 18 (3). – P. 207-212.
28. Ziomek A., Oszust M. Evaluation of Interest Point Detectors in Presence of Noise // Interna-tional Journal of Intelligent Systems and Applications (IJISA). – 2016. – Vol. 8, No. 3. – P. 26-33.
29. Calonder M., Lepetit V., Strecha C., Fua P. BRIEF: Binary robust independent elementary features, in: ECCV, 2010.
30. Leutenegger S., Chli M., Siegwart R.Y. BRISK: Binary Robust invariant scalable keypoints // 2011 International Conference on Computer Vision. – 2011. – P. 2548-2555.
31. Rublee E., Rabaud V., Konolige K., Bradski G.R. ORB: An efficient alternative to SIFT or SURF. ICCV. – 2011. – P. 2564-2571.
32. Cowan B., Imanberdiyev N., Fu C., Dong Y., Kayacan E. A performance evaluation of detec-tors and descriptors for UAV visual tracking // Proc. Int. Conf. Control, Autom., Robot. Vis. (ICARCV). – 2016. – P. 1-6.
33. Fischler M.A., Bolles R.C. Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography // Comm. ACM. – 1981. – Vol. 24 (6). – P. 381-395.
Published
2019-05-08
Issue
No 1 (2019)
Section
SECTION V. TECHNICAL VISION