Title :
Image scale and rotation from the phase-only bispectrum
Author_Institution :
Dept. of Electr. & Information Eng., Oulu Univ., Finland
Abstract :
This paper deals with the problem of aligning two images under translation, rotation and sealing. The method described utilizes the shift invariance property of the bispectrum to eliminate the effect of the translation component. Only the phase information is preserved from the bispectrum in order to achieve better resilience against nonuniform illumination changes. The scale and the rotation parameters are estimated from the remaining log-polar sampled spectrum using cross-correlation. The examples shown in the paper indicate that the method is quite robust against background clutter and occlusions.
Keywords :
clutter; correlation methods; hidden feature removal; image processing; parameter estimation; transforms; background clutter; image scale; log-polar sampled spectrum; nonuniform illumination change; parameter estimation; phase-only bispectrum; shift invariance property; translation component; Data mining; Fourier transforms; Image segmentation; Lighting; Machine vision; Parameter estimation; Pattern recognition; Resilience; Robustness; Voting;
Conference_Titel :
Image Processing, 2004. ICIP '04. 2004 International Conference on
Print_ISBN :
0-7803-8554-3
DOI :
10.1109/ICIP.2004.1421420