-
8/3/2019 Ifsr Introduction
1/14
Image Registration:
Introduction
Arthur Goshtasby
ri ht t te ni ersit
Image Registration and Fusion Systems
-
8/3/2019 Ifsr Introduction
2/14
s an array o sca ars or vec ors.
Scalar:Reflectance, temperature, range
ec or: , mu spec ra , yperspec ra
Landsat MSS image,
courtes of NASA
A digital image
2
-
8/3/2019 Ifsr Introduction
3/14
images of a scene. The process in effect
establishespoint-by-point correspondence between the images.
Landsat MSS Landsat TM Registered MSS & TM
3
Data courtesy of NASA
-
8/3/2019 Ifsr Introduction
4/14
Reference or source ima e: This is the
image that is kept unchanged and is usedas the reference.
image that is resampled to spatially align
with the reference image. rans orma on unc on: e unct on
that is used to resample the target imageto the geometry of the
reference image.
Control points: Unique landmarks in theimages. Corresponding
control points are
4
parameters.
-
8/3/2019 Ifsr Introduction
5/14
1. Change detection (intensity)
Landsat 1 Landsat 2 Change image
Data courtesy of NASA
5
-
8/3/2019 Ifsr Introduction
6/14
2. Change detection (geometry)
(a), (b) Temporal MR brain images. (c), (d) Before and after
registration.
6
.
-
8/3/2019 Ifsr Introduction
7/14
.
Temporal brain imagesImage flow of one image slice
7
Data courtesy of Kettering Medical Center
-
8/3/2019 Ifsr Introduction
8/14
4. Image fusion
Data courtesy of NASA
8
Landsat TM bands 1 & 7
-
8/3/2019 Ifsr Introduction
9/14
5. Image mosaicking
Mosaicked image
Two aerial images of Honolulu, HI.
9
-
8/3/2019 Ifsr Introduction
10/14
6. Depth perception
Data courtesy of Tsukuba University
10
A pair of stereo images
-
8/3/2019 Ifsr Introduction
11/14
. Target localization
Target recognition
Target tracking Scene 3-D recovery
11
-
8/3/2019 Ifsr Introduction
12/14
deblurring, image segmentation, edgedetection
2. Feature extraction and selection:
Extractin oints, lines, re ions, tem lates,etc. from images and
selecting some
3. Correspondence: Determining the
correspondence between selected features inimages
12
-
8/3/2019 Ifsr Introduction
13/14
.From the corresponding feature points
5. Resampling: Resampling the sensed image
to t e geometry o t e re erence mage us ng
the transformation
13
-
8/3/2019 Ifsr Introduction
14/14
Books and special issues on
image registration1. Image Registration for Earth Science,
Jacqueline Le Moigne,
at an etanya u, an oger astman s. , am r ge ress,
2010.2. Landmark-Based Image Analysis: Using Geometric and
Intensity
, . , , , ,2001.
3. 2-D and 3-D Image Registration, A. Goshtasby, John Wiley
&Sons, New York, NY, Feb. 2005.
4. Non-Rigid Image Registration, A. Goshtasby, L. Staib,
C.Studholme, and D. Terzopoulos (Eds.), a special issue ofComputer
Vision and Image Understanding, Feb. 2003.
. mage eg strat on, . os tas y, . e o gne s. , a spec aissue
ofPattern Recognition, Jan. 1999.
6. A survey of image registration techniques, Lisa Brown,ACM
14
, .
