Abstract: According to the textural characteristics of macerals in exinite of coal and the difference between them, features of coarseness, contrast, directionality, linelikeness and regularity of samples were extracted with Tamura texture description method, which is based on human visual apperception, to describe the typical three macerals of cutinite, alginite and barkinite, and the distinguishability of features were analyzed. In view of the limitations of the sample number of exinite macerals, a support vector machine based classifier, which is fit for small sample size problem was built, features (single or joint) were employed for the classification of the three typical macerals.Experimental results show that it is difficult to achieve an effective classification with single feature, but after joint two of them, the effect of classification is improved obviously, and the combination of linelikeness and directionality performs its best, achieves an accuracy of 98.9%, which can offer a reference for further classification or recognition of other macerals of coal.