Feature points or landmarks are unique points in an image, characterizing the geometric layout of an image.
Landmarks are most useful because their coordinates can be directly used to determine the parameters of
a transformation function to register two images. Examples of landmarks are given below. Various landmark
selection algorithms have been developed by Image Registration and Fusion Systems. With these
algorithms, one can adjust the spacing between the landmarks and select a desired number of them in an
image. Landmarks that are stable under a wide range of image resolutions and orientations are selected.
The invariance property of landmarks makes it possible to select the same landmarks in images captured
under different environmental conditions or different camera settings.
Two examples of landmark selection are given below. For more examples, visit the Point detectors page.
Fig. 1. (top) A desired number of landmarks selected in two images of the Art Museum in Washington, D.C., taken from
slightly different views. Many of the landmarks show the same physical points in the scene. (bottom) Landmarks selected in
two images of the World Trade Center, New York City. Original images of the New York City are courtesy of NASA.
To obtain a license for a software that selects a
desired number of landmarks in an image,
follow this link =>
Point (landmark) selection
|Image Registration and Fusion Systems
Each software tool comes in a C/C++ static library,
compiled under Microsoft Visual Studio under
Windows OS. The library comes with a C/C++ driver
program demonstrating usage of the software.