CE HARALICK - Contrast Enhancement - Haralick

(6/16/01)

PURPOSE

Texture based segmentation of image using Haralick operators.   Example.

SEE ALSO

CE VAR [Contrast Enhancement - Variance]
CE HURST [Contrast Enhancement - Hurst]

USAGE

.OPERATION: CE HARALICK

.INPUT FILE: IMG001
[Enter the name of the input file.]

.OUTPUT FILE: IMG002
[Enter name for output file.]

.NEIGHBORHOOD X & Y: 13 13
[Neighborhood around a pixel. Must be odd numbers.]

.INTENSITY LEVELS: 32
[Intensity levels for co-occurance matrix.]

.OFFSET IN X & Y: 2 2
[Offsets for co-occurance.]

.MODE NUMBER (1...6): 1
[Mode 1: Homogeneity using second moment
Mode 2: Contrast using difference moment.
Mode 3: Weighted average absolute distance from diagonal. Gose, Johnsonbaugh & Jost, "Pattern Recognition and Image Analysis" refer to this as "inertia".
Mode 4: Castleman refers to this as "entropy".
Mode 5: Most probable intensity at the offset.
Mode 6: Linear dependency of brightness. ]

NOTES

  1. Image is "circularly closed" so that pixels on boundaries have a neighbor on opposite boundary.

  2. Uses Co-occurance of brightness differences inside neighborhood. Image is mapped to a limited number of intensity levels before finding the co-occurance matrix.

  3. Reference: John C. Russ, "The Image Processing Handbook", CRC Press, Inc. (2000)

  4. I have not implemented this for volumes since most EM volumes do not have a strong anisotropy in textures with direction, thus a 2D approach is just as useful as 3D. Volumes are processed slice-by-slice.

SUBROUTINES

FILTER_HAR

CALLER

UTIL2

© Copyright Notice /       Enquiries: spider@wadsworth.org