COOCCURRENCE MATRICES FOR SMALL REGION TEXTURE MEASUREMENT AND COMPARISON

被引：7

作者：

CARLSON, GE ^{[1
]}

EBEL, WJ ^{[1
]}

机构：

[1] MISSISSIPPI STATE UNIV,DEPT ELECT ENGN,STARKVILLE,MS 39762

来源：

INTERNATIONAL JOURNAL OF REMOTE SENSING | 1995年 / 16卷 / 08期

关键词：

D O I：

10.1080/01431169508954485

中图分类号：

TP7 [遥感技术];

学科分类号：

081102 ; 0816 ; 081602 ; 083002 ; 1404 ;

摘要：

Modification of the typical co-occurrence matrix measure of image region texture is discussed. The modification produces a co-occurrence array with elements that represent non-directional joint grey level probabilities. The array size varies with the number of pixel adjacencies available in the image region. For small regions this gives better joint probability estimates and thus produces more effective and consistent texture comparisons for small regions. A simple segmentation example is given to illustrate the techniques described.

引用

页码：1417 / 1423

页数：7