Radiotekhnika
Publishing house Radiotekhnika

"Publishing house Radiotekhnika":
scientific and technical literature.
Books and journals of publishing houses: IPRZHR, RS-PRESS, SCIENCE-PRESS


Тел.: +7 (495) 625-9241

 

Image preprocessing for stochastic gradient estimation of inter-frame geometric deformations

Keywords:

P.V. Yakshankin – Post-graduate Student of Ulyanovsk State Technical University. E-mail: yakspavel@yandex.ru
I.V. Voronov – Post-graduate Student of Ulyanovsk State Technical University. E-mail: ilvo1987@gmail.com


When estimating the parameters of inter-frame geometric deformations of images stochastic gradient estimation algorithms are a good choice due to their high speed and accuracy. However, these algorithms have a small operating range of parameters to be estimated. The image preprocessing procedure increasing the inter-frame geometric deformation parameters’ domain of stochastic gradient estimation procedure is proposed and investigated. The preprocessing includes feature detection and Gaussian filtering. Feature points are identified using a modified FAST detector. Each matching point is associated with confidence factor using as a parameter of th following a Gaussian filter. Examined preprocessing of images can considerably expand the effective range of the stochastic gradient algorithm. SGA with preprocessing is more sensitive to noise due to FAST detector using raw intensities. The use of only feature points with applying Gaussian filter influences the accuracy of estimation. However, SGA with preprocessing can be used as a good initial guess for further estimation with higher accuracy.
References:

  1. Tashlinskii A.G. Computational expenditure reduction in pseudo-gradient image parameter estimation // Lecture Notes in Computer Science. 2003. V. 2658. P. 456-462.
  2. Tashlinskij A.G. Psevdogradientnoe oczenivanie prostranstvenny'x deformaczij posledovatel'nosti izobrazhenij // Naukoemkie texnologii. 2002. T. 3. S. 32-43.
  3. Vasil'ev K.K., Krasheninnikov V.R., Tashlinskij A.G. Statisticheskij analiz posledovatel'nostej mnogomerny'x izobrazhenij // Naukoemkie texnologii. 2013. T. 14. № 5. S. 5-11.
  4. Rodzhers D. Algoritmicheskie osnovy' mashinnoj grafiki. M.: Mir. 1989. 512 c.
  5. Gonsales R., Vuds R. Cifrovaya obrabotka izobrazhenij. M.: Texnosfera. 2005. 1072 s.
  6. Rosten E., Drummond T. Fusing points and lines for high performance tracking // IEEE International Conference on Computer Vision. 2005. V. 2. P. 1508–1511.

© Издательство «РАДИОТЕХНИКА», 2004-2017            Тел.: (495) 625-9241                   Designed by [SWAP]Studio