Method of ground-objects isolation against the background of underlying surface in on-board 3d laser channel

Electronics, Radio and Communications


Аuthors

Mirzoyan A. S.1*, Khmarov I. М.2**, Kondrashov N. G.3***, Shakhov S. V.4****

1. Rybinsk State Aviation Technical University named after P.A. Soloviev, RSATU, 53, Pushkin St., Rybinsk, Yaroslavl region, 152934, Russia
2. Research center (Tver) of Central Research Airforce Institute of the Russian Defense Ministry, 32, Afanasy Nikitin emb., Tver, 170026, Russia
3. Air force academy named after professor N.E. Zhukovskii and Y.A. Gagarin, Voronezh, Russia
4. Tver State Technical University (Tver), TSTU,, 22, Naberezhnaya Afanasii Nikitin, Tver, 170026, Russia

*e-mail: andr.s.mirzoyan@gmail.com
**e-mail: khmarov314@mail.ru
***e-mail: nik-avia@mail.ru
****e-mail: Svshahov@yandex.ru

Abstract

Application of aircraft laser optic-electronic channel for 3D ground-objects image acquisition is the perspective direction of automatic recognition system development. For secure ground-objects recognition their qualitative isolation against the background of underlying surface is required. Ground-objects isolation is suggested to realize on the basis of 2D underlying surface approximation of in laser channel coordinate system.
The algorithm of underlying surface approximation by principal components method is presented in the work. The result of algorithm work is the surface carrying the best mean-square underlying surface approximation. Obtained 2D underlying surface approximation in 3D space makes possible to isolate ground-object against the background of underlying surface and to make its 3D portrait by the heights matrix from the approximating surface. Advantage of 3D portrait by the heights matrix over 3D portrait by the slant distance matrix consists in its invariance in the wide range of observation angle change and the possibility of removing «smearing» effect by the methods based on the appropriate transformation of spatial triangles that make up the outer boundary of 3D portrait.
Recognition of ground-objects against the background of underlying surface is supposed to carry out on the basis of geometrical characteristics of their 3D portraits by the heights matrix. The main geometrical characteristics are the object projection area on the underlying surface, convex hull projection area, projection diameter etc.
Efficiency of the suggested algorithms is checked on the irregular geometrical shape ground-objects images acquired as a result of laser optic-electronic channel work modeling. Testing is carried out on the laser images of different quality by essential variations of observation angle, angle of elevation, object distance, object orientation in the space, and arrangement of its links relative to each other. Due to the sufficient quantity of numerical experiments its determined that declinations of the main geometrical characteristics of 3D portraits by slant distance matrix from the nominal values are within the limits that make possible to realize the algorithms of automatic ground-objects recognition by their laser images. Keywords: two dimensional approximation, methodof principal components, laser channel.

Keywords:

two dimensional approximation, method of principal components, laser channel

References

  1. Ping Yuan, Rengarajan Sudharsanan, Xiaogang Bai, Joseph Boisvert, Paul McDonald, Eduardo Labios, Michael S. Salisbury, Gary M. Stuart, Harrison Danny, Angel A. Portillo, Alric B. Roybal, Stephen Van Duyne, Greg Pauls, Steve Gaalema.32 ? 32Geiger-mode ladar camera, Proceedings ofthe SPIE, 2010, vol.7684.
  2. Christophe Harvey, Jonathan Wood, Peter Randall, Graham Watson, Gordon Smith. Simulation ofanew 3Dimaging sensor for identifying difficult military targets, Proceedings ofthe SPIE, 2008, vol.6950.
  3. Angela M. Kim, Richard C. Olsen, Carlos F. Borges Simulating full-waveform lidar, Proceedings ofthe SPIE, 2010, vol.7684.
  4. Jesse L. Hodge, David H. DeKruger, Alden E. Park. Mobile target ladar ATR system, Proceedings ofthe SPIE, 2001, vol.4379, , pp.35-50.
  5. Alexandru N. Vasile, Richard M. Marino. Pose Independent Automatic Target Detection and Recognition using 3-D Laser Radar Imagery, Lincoln laboratory journal, 2005, vol.15, no.1, pp. 61-78.
  6. Nieves RubenD., Reynolds William D. Jr. Three- dimensional transformation for automatic target recognition using lidar data, Proceedings ofthe SPIE, 2010, vol.7684.
  7. Khmarov I.M., Kanivets V.Yu., Kondrashov N.G. Uspekhi sovremennoi radioelektroniki, 2010, no.11, pp. 86-90.
  8. Pearson, K.Onlines and planes ofclosest fit tosystems ofpoints inspace. Philosophical Magazine, ser. VI2, pp. 559-572 (1901).
  9. Gorban A.N., KeglB., Wunsch D.C., Zinovyev A. Principal Manifolds for Data Visualization and Dimension Reduction. Springer. Series: Lecture Notes inComputational Science and Engineering, vol.58, 2008, 340p.
  10. Polyakov V.M., Pokrovskii V.P., Soms L.N. Opticheskii zhurnal, 2011, vol.78, no.10, pp. 4-9. 
  11. Mirzoyan A.S. Vestnik Rybinskogo gosudarstvennogo aviatsionnogo tekhnicheskogo universitetaim. P.A. Soloveva, 2012, no.2, pp. 186-192

mai.ru — informational site of MAI

Copyright © 1994-2024 by MAI