manpagez: man pages & more
man cooc(1)
Home | html | info | man
cooc(1)                                                                cooc(1)


       cooc,  cooc_features  - calculate the co-occurrence matrix and features
       on it


       cooc image matrix xpos ypos xsize ysize dx dy flag

       cooc_features matrix


       cooc creates a 256 by 256 one channel co-occurrence matrix of  the  box
       determined  by  the  parameters (xp, yp; xs, ys) within the image file.
       The matrix is written onto the Vasari image file matrix.  The displace-
       ment vector is determined by (dx, dy).  The user must ensure that there
       is enough border pixels around the box within im dictated by  the  dis-
       placement vector (dx,dy) or else the program fails.  All entries of the
       co-occurrence matrix are double  normalised  to  the  number  of  pairs
       involved.   This function is a direct implementation of the paper: Har-
       alick R. M., Shanmugan K. and Dinstein I., 'Textural features for image
       classification',  IEEE  Transactions  on Systems, Man, and Cybernetics,
       Vol. SMC-3, No 6, Nov. 1973, pp 610-621.

       If flag sym is 1, the created co-occurrence matrix is symmetric that is
       dispacement  vectors  (dx,  dy),  (-dx,  -dy)  create  exactly the same
       matrix.  If sym is 0, the created co-occurrence matrix is not symmetric
       that  is  dispacement  vectors  (dx,  dy),  (-dx, -dy) create different

       Input image should be one band unsigned char image.

       cooc_features calculates and prints at the standard error  output  fea-
       tures of the cooccurrence matrix matrix.


       im_glds_matrix(3X), im_cooc_asm(3X), im_cooc_contrast(3X), im_cooc_cor-
       relation(3), im_cooc_entropy(3X)


       N. Dessipris


       N. Dessipris - 27/2/1992

                                  27 Jan 1992                          cooc(1)

cooc 7.14.5 - Generated Fri Sep 5 18:41:53 CDT 2008
© 2000-2021
Individual documents may contain additional copyright information.