This paper develops a technique for the registration of multisensor images utilizing the Laplacian of Gaussian (LoG) filter to automatically determine semi-invariant ground control points (GCPs). These points are then related through the development of point matching techniques and statistical analysis. Through the use of matrix transformations, efficient management of multiple affine operations can be obtained and stored in a composite transform. Wavelet theory is used to enable the multi-resolution analysis critical for multisensor image registration and predictive transformations. Multiple methods have been discussed to test the accuracy of the resulting image registration. Benefits of this technique against parallax and moving objects within the scene has also been highlighted. Finally, an example of 'wavelet sharpening' has been demonstrated that preserves radiometric integrity.
[1]
George Wolberg,et al.
Digital image warping
,
1990
.
[2]
Karl C Walli,et al.
Multisensor Image Registration Utilizing The LoG Filter and FWT
,
2003
.
[3]
Adith Chandrasekhar.
Point extraction and matching for registration of infrared astronomical images
,
1999
.
[4]
Robert A. Schowengerdt,et al.
Remote sensing, models, and methods for image processing
,
1997
.
[5]
John R. Schott,et al.
Remote Sensing: The Image Chain Approach
,
1996
.
[6]
Stéphane Mallat,et al.
A Theory for Multiresolution Signal Decomposition: The Wavelet Representation
,
1989,
IEEE Trans. Pattern Anal. Mach. Intell..
[7]
Laurent Peytavin,et al.
Cross-sensor resolution enhancement of hyperspectral images using wavelet decomposition
,
1996,
Defense + Commercial Sensing.