Conclusion#
We have seen in that chapter that methods for finding features are very different and depend on the seeking feature.
For edges, usual methods are essentially a filtering using the gradient or the Laplacian (Roberts or Prewitt filters, Sobel or Canny detectors).
For corners, the dedicated methods analyze the variation in intensity in the neighbourhood of the pixels (Moravec and Harris detectors).
For lines and circles, the image is represented in the space of the parameters of the geometric shape (Hough transform).
References#
J. Canny, “A Computational Approach To Edge Detection”, IEEE Transactions on Pattern Analysis and Machine Intelligence}, vol.8, 1986.
C. Harris and M. Stephens “A combined corner and edge detector”, Alvey Vision Conference, p. 147-151, 1988.
P.V.C. Hough, “Method and means for recognizing complex patterns”, US Patent 3,069,654, 1962.
D. Marr and E. Hildreth, “Theory of Edge Detection”, Proceedings of the Royal Society of London, vol. 207, 1980.
H. Moravec, “Obstacle Avoidance and Navigation in the Real World by a Seeing Robot Rover”, technical report, Carnegie–Mellon University, Robotics Institute, 1980.
J.M.S. Prewitt, “Object enhancement and extraction”, Picture Processing and Psychopictorics, Academic Press, 1970.
L.G. Roberts, “Machine Perception Of Three-Dimensional Solids”, Computer Methods in Image Analysis, IEEE Press, 1965.
I. Sobel and G. Feldman, “A \(3\times3\) Isotropic Gradient Operator for Image Processing”, In Stanford Artificial Intelligence Project, 1968.