jdm5548@tamsun.tamu.edu (James Darrell McCauley) (03/20/91)
I'm not a comp.graphics regular, or even a graphics person, so bear with me. I'm trying to write code to find textural features, a la "Textural Features for Image Classification" by R.M. Haralick, K. Shanmugam, and I. Dinstein (IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS, v. SMC-3, No. 6, Nov. 1973). (I'm just starting to learn about texture...) They describe 14 textural features, some of which are: angular second moment, contrast, correlation, sum of squares (variance), entropy. These all give a single numeric value describing an image. To get these features, is necessary to find what they call a "gray-tone spatial dependence matrix" (this paper's kind of old, so I'm unsure of the terminology here). To quote: "Such matrices of gray-tone spatial-dependence frequencies are a function of the angular relationship between the neighboring resolution cells as well as a function of the distance between them." Now, I'm hoping that someone knows what I/they mean by "gray-tone spatial dependence matrix," cause here's the big question: Does anyone have an algorithm any code to find this matrix given any angle that is a factor of pi/4 and any distance d? I'm attempting to find the 14 textural features for images in PGM format (as in PBMPLUS). I started writing code to find this matrix, but the farther I got, the uglier (and longer) it got. Has anyone already done this? Thanks, -- James Darrell McCauley (jdm5548@diamond.tamu.edu, jdm5548@tamagen.bitnet) Spatial Analysis Lab, Department of Agricultural Engineering, Texas A&M University, College Station, Texas 77843-2117, USA