This matrix is a tabulation of how often different combinations of pixel
brightness values occur in an image according to a particular
displacement vector that shows the direction towards we look for the
occurrences. We used a GLCM normalization that expresses the probability
of occurrence.
Each Frame is divided into Macroblocks (16x16 pixels).
Each GLCM is computed according to a displacement vector
d.
GLCM is computed for each Macroblock.
Different statistics are calculated and placed into the feature vectors.
These vectors are the inputs for the segmentation algorithm
•4
main directions have been used so as to compute the occurrences:
0°
or (1,0);
45°
or (1,1);
90°
or (0,1);
135°
or (-1,1).
Example of a GLCM matrix for a 4x4-pixel Macroblock, using d=1 and
direction 0º.