Avtor/Urednik		Pušenjak, Jani; Kotnik, Tadej; Likar, Boštjan; Pernuš, Franjo
Naslov		Določanje zamika med slikama s kepstralno metodo
Prevedeni naslov		Image shift registration by the cepstrum technique
Tip		članek
Vir		In: Zajc B, editor. Zbornik 7. elektrotehniške in računalniške konference ERK'98. Zvezek B. Računalništvo in informatika, umetna inteligenca, robotika, razpoznavanje vzorcev, biomedicinska tehnika, močnostna elektrotehnika, didaktika, študentski članki; 1998 sep 24-26; Portorož. Ljubljana: Slovenska sekcija IEEE,
Leto izdaje		1998
Obseg		str. 343-6
Jezik		slo
Abstrakt		Registration of biological and physiological processes is often based on a comparison between two images taken at different times. In biomedical imaging, it is practically impossible to acquire two images with the observed object having the same relative position, rotation, magnification and intensity at the same time. To compare the two images, these differences have first to be diminished. This paper deals with the cepstrum technique for solving the problem of translational shift between two images of the same object. This method is based on the calculation of the two-dimensional power cepstrum of the cepstral window constructed from the two images positioned one beside the other. For two images, each of size DxD, and the object relatively shifted by (x0. y0) in one of the images with respect to the other, the power cepstrum of the cepstral window has a distinguishable peak at (Ix0I, sign(D/2 - 1 y0)). This peak does not uniformly define the shift, but gives two possible solutions. The correct one has to be pinpointed either manually by the observer, which is feasible with larger shifts, or by an automated method. The cepstrum technique is very robust regarding the noise and the differences in the intensity of the images. it is, however, sensitive to rotation. Therefore, larger rotaions have to be determined and eliminated before applying the technique for determination of the translational shift.
Deskriptorji		IMAGE INTERPRETATION, COMPUTER-ASSISTED IMAGE PROCESSING, COMPUTER-ASSISTED