The Hough transform is useful for detecting objects of known shape (i.e. objects that can be described mathematically by specific parameters), but of unknown location. Furthermore, Hough techniques are relatively unaffected by noise or missing data in images.
Detecting a straight line in a noisy image may appear to be of limited value. Many man-made objects however contain linear features, thus, locating man-made objects in images often requires locating straight lines.
One application for which Hough techniques are suitable is that of locating
man-made objects in aerial photographs.
The Hough technique employed in locating straight lines can be generalized
to detect shapes of higher complexity such as circles and ellipses.
This gives rise to a range of applications for which the generalized
Hough transfrom can be employed. Examples include that of detecting
tumors in chest film, and detecting hemoglobin fingerprints [1].