| Landmark correspondence |
| Image Registration and Fusion Systems |
|
|
Landmark correspondence or point pattern matching is the process of finding correspondence between landmarks in two
images of a scene. If landmarks in only one of the images are given, corresponding landmarks in the second image can be
determined through a process known as template matching. If landmarks in both images are given, a point pattern matching
algorithm is required to determine the correspondence between the points.
An example of landmark correspondence where landmarks in both images are given is provided below. Images in the top row
show the landmarks (by red dots) detected in two Landsat 5 images of a scene. The scene is relatively flat and because the
sensor is very far from the scene compared to variations in scene elevation, corresponding landmarks in the images can be
related by an affine transformation. The landmarks in the images are detected by a procedure described elsewhere (See the
landmark detection page for details).
Although many landmarks are provided in the images, due to considerable changes in the scene between the times the
images were obtained, there are many landmarks that exist in only one of the images. The algorithm is capable of finding the
correspondence between landmarks that appear in both images and register the images. Registration of the images by an
affine transformation (see the auto affine registration page for details) using the obtained set of correspondences is shown in
the left image in the bottom row. Subtracting the registered images we obtain the right image in the bottom row, showing
scene changes occurring between the times the images were obtained. This software assumes that landmarks in the images
are related by an affine transformation.
images of a scene. If landmarks in only one of the images are given, corresponding landmarks in the second image can be
determined through a process known as template matching. If landmarks in both images are given, a point pattern matching
algorithm is required to determine the correspondence between the points.
An example of landmark correspondence where landmarks in both images are given is provided below. Images in the top row
show the landmarks (by red dots) detected in two Landsat 5 images of a scene. The scene is relatively flat and because the
sensor is very far from the scene compared to variations in scene elevation, corresponding landmarks in the images can be
related by an affine transformation. The landmarks in the images are detected by a procedure described elsewhere (See the
landmark detection page for details).
Although many landmarks are provided in the images, due to considerable changes in the scene between the times the
images were obtained, there are many landmarks that exist in only one of the images. The algorithm is capable of finding the
correspondence between landmarks that appear in both images and register the images. Registration of the images by an
affine transformation (see the auto affine registration page for details) using the obtained set of correspondences is shown in
the left image in the bottom row. Subtracting the registered images we obtain the right image in the bottom row, showing
scene changes occurring between the times the images were obtained. This software assumes that landmarks in the images
are related by an affine transformation.
To obtain a license for this software, please follow this link =>
Fig. 1. (Top row) Temporal Landsat images. These images are courtesy of NASA. (Bottom left) Registered images. (Bottom right)
Absolute difference of registered images.
Absolute difference of registered images.
This software comes in a C/C++ static library, compiled with
Microsoft Visual Studio under Windows PC. The library comes
with a C/C++ driver program demonstrating usage of the software.
Microsoft Visual Studio under Windows PC. The library comes
with a C/C++ driver program demonstrating usage of the software.
