M.B. Nikiforov – Ph. D. (Eng.), Associate Professor, RSREU (Ryazan). E-mail: email@example.com
A.I. Novikov – Ph. D. (Econ.), Associate Professor, RSREU (Ryazan). E-mail: firstname.lastname@example.org
C.V. Orlov – Undergraduate, RSREU (Ryazan). E-mail: email@example.com
The problem of creating an intelligent technical vision systems working in real time in onboard aircraft system, has a high level of relevance. In this issue one of the most important objective is to build a 3D image of the underlying surface based on the images of the object obtained from a sensor (video camera, thermal imager, etc.). The algorithm processes a set of several images or a video stream. Knowing the coordinates of a particular point of the same object on two (or more) images obtained from different angles of view allow to calculate the position of a point in three-dimensional space (if you know parameters of the cameras).
The three-dimensional reconstruction task consists of several steps, which are an important part of the problem:
finding the key (singular) points (building of detectors);
building of key points descriptions (descriptors) and establishment of correspondences between particular points on pair of the images.
In the first part of the main section of the present work summarizes the major steps of the SURF method, which used to find a key points. This work is devoted to building of the descriptors, which have low computing complexity and show a relatively good accuracy characteristics. Base of the new method is the assumption the surface in a neighborhood of a key point defined by brightness function adequately describes the algebraic function of low (2nd, 3rd) degree. The surface defined by this function is broken down into 4 or 8 sectors, each of which calculates the normal vector to the corresponding part of the surface. Authors provide detailed algorithms and examples of normal vectors calculating, and the results of applying this method to establish correspondence between pairs of points on the two real images.
Herbert Bay, Andreas Ess,
Tinne Tuytelaars, Luc Van Gool Speeded-Up Robust Features (SURF). Elsevier. CH-8092 Zurich. B‑3001 Leuven. 2008. 14 p.
Prehtt U. Cifrovaja obrabotka
izobrazhenijj / Per. s angl. M.: Mir. 1982. Kn. 2. 480 s.
Lowe D.G. Distinctive Image Features from Scale-Invariant
Keypoints // International Journal of Computer Vision. 2004. V. 2. № 60. P. 91−110.
Novikov A.I., Sablina V.A.,
Nikiforov M.B., Loginov A.A.
The Contour Analysis and Image-Superimposition Problem in Computer Vision Systems
// Pattern Recognition and Image Analysis. 2015. V. 25. № 1. P. 73−80.
Harris S., Stephens M.
A combined corner and edge detector // Proc. of
4th Alvey Vision Conference. 1988. P. 147−151.
Elesina S.I., Kostjashkin L.N.,
Loginov A.A., Nikiforov M.B.
Sovmeshhenie izobrazhenijj v korreljacionno-ehkstremalnykh navigacionnykh sistemakh.
M.: Radiotekhnika. 2015. 208 s.
Loginov A.A., Muratov E.R.,
Nikiforov M.B., Novicov A.I.
Reducing the computational complexity of image registration in the aircraft technical
vision systems // Dynamics of complex systems. 2015. V. 9. № 1. P. 39−45.