Edge Detection of Multispectral Images Using the 1-D Self-Organizing Map

In this paper, a new method for edge detection in multispectral images is presented. It is based on the use of the Self-Organizing Map (SOM) and a conventional edge detector. The method presented in this paper orders the vectors of the original image in such a way that vectors that are near each other according to some similarity criterium should have scalar ordering values near each other. This is achieved using the 1-dimensional Self-Organizing Map. After ordering, the original vector image reduces to a gray-value image, and conventional edge detectors can be applied. In this paper, the Laplace and Canny edge detectors are used. It is shown, that using the Self-Organizing Map (SOM) in ordering the vectors of the original spectral image it is possible to find also those edges that the R-ordering based methods miss.

[1]  Douglas R. Anderson,et al.  Mapping Multidimensional Space to One Dimension for Computer Output Display , 1968, IEEE Transactions on Computers.

[2]  Aldo Cumani,et al.  Edge detection in multispectral images , 1991, CVGIP Graph. Model. Image Process..

[3]  Panos E. Trahanias,et al.  Color edge detection using vector order statistics , 1993, IEEE Trans. Image Process..

[4]  John F. Canny,et al.  A Computational Approach to Edge Detection , 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[5]  J. Astola,et al.  Vector median filters , 1990, Proc. IEEE.

[6]  V. Barnett The Ordering of Multivariate Data , 1976 .

[7]  Silvano Di Zenzo,et al.  A note on the gradient of a multi-image , 1986, Comput. Vis. Graph. Image Process..