Image registration using wavelet techniques

Harold S. Stone; Jacqueline Le Moigne; Morgan McGuire

doi:10.1117/12.300048

1 March 1998 Image registration using wavelet techniques

Harold S. Stone, Jacqueline Le Moigne, Morgan McGuire

Proceedings Volume 3240, 26th AIPR Workshop: Exploiting New Image Sources and Sensors; (1998) https://doi.org/10.1117/12.300048
Event: 26th AIPR Workshop: Exploiting New Image Sources and Sensors, 1997, Washington, DC, United States

Abstract

The objective of this paper is to do fast image registration by progressively registering wavelet representations at different resolutions. It is well known that some subbands of wavelet representations are sensitive to small translations of an image. The energy in some subbands of wavelet representations are sensitive to small translations of an image. The energy in some subbands can move entirely out of the subband into other subbands when forming wavelet representations of shifted versions of an image. This seems to preclude the practical use of wavelet representations for image registration. In this paper we show that registration with wavelet techniques can be done effectively, and show this by examining the sensitivity of wavelet representations to image translations. Experiments on general mathematical modes and actual satellite data produced these findings: 1. Registration using a wavelet with a block size of B pixels is robust under image translation for features that extend at least 2B pixels. Distortion from translation causes the peak correlation in the low-pass subband to vary only between roughly 0.8 and 1.0, with an average in excess of 0.9 for features larger than 2B. 2. The high-pass band is less robust than the low-pass subband. Edge information at low resolution can best be used by using the high-pass subband of a low-resolution image created by recursive low- pass filtering of the original image. For this case, translation distortion drops the correlation to a range from 0.2 to 0.8, which is low in some cases but still useful on average for registration. 3. There is little difference in correlations produced from Haar and Daubechies wavelet transforms. 4. The mathematical model and the experiments with areal image gave consistent results.

Citation Download Citation

Harold S. Stone, Jacqueline Le Moigne, and Morgan McGuire "Image registration using wavelet techniques", Proc. SPIE 3240, 26th AIPR Workshop: Exploiting New Image Sources and Sensors, (1 March 1998); https://doi.org/10.1117/12.300048

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available