关键词: 微生物, 显微图像, 纹理, 变差函数, 支持向量机

Abstract: Because bacterium in microbe image has different shapes, is apt to overlap, and gray levels are very close, a new method of micrograph identification was proposed in this paper, which was based on pattern recognition theory of Support Vector Machine (SVM) and used texture feature extracted by Variation. The classification results were then compared with the results obtained by NN theory. The results of study have proved that the method based on pattern recognition theory of support vector machine to the classification of bacterium micrograph may improve the accuracy of image classification.

Key words: bacterium, micrograph, texture, variation, Support Vector Machine (SVM)