BibTeX
@MISC{Flusser_registrationof,
author = {Jan Flusser and Jirí Boldys},
title = {Registration of N-D Images by Blur Invariants},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
. Registration of monomodal as well as multimodal images is a fundamental task in medical image processing. In this paper, the attention is paid to landmark matching. We present a new class of invariants dened on a volumetric neighborhood of a landmark. They are invariant to convolving the image by arbitrary centrosymmetric function and to image translation. Since the method operates directly on the image grey values, it is particularly designed for multitemporal N-D images of the same modality. The images are assumed to be degraded by an unknown amount of blur, introduced by the point-spread function of the imaging system. 1 Introduction Registration of monomodal as well as multimodal images is a fundamental task in medical image processing [23]. Regardless of the modalities involved and of the image dimensions, the registration usually consists of the four following steps. Landmark detection. In medical imaging, landmarks can be signicant anatomical points easily locata...
Keyphrases
blur invariant n-d image medical image processing fundamental task multimodal image landmark detection point-spread function image dimension new class volumetric neighborhood arbitrary centrosymmetric function introduction registration medical imaging signicant anatomical point imaging system unknown amount image translation image grey value multitemporal n-d image
