基于小波变换的工业微生物菌种自动识别  

Research for Industrial Microbial Germ Automatic Recognition Based on Wavelet Transformation

作　　者：赵增华[1] 舒炎泰[1] 王积分[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《天津大学学报（自然科学与工程技术版）》2002年第2期151-154,共4页Journal of Tianjin University：Science and Technology

基　　金：国家自然科学基金资助项目 (696750 0 6).

摘　　要：以“绿僵菌”为例 ,在前人研究的基础上 ,从包含原图像丰富信息的多分辨率小波分解的子图像抽取纹理熵特征 ,经特征选择 ,确定最佳特征组合维数 ,设计了基于感知准则的线性分类器 ,训练和测试后 ,分类识别精度可达 90 %以上 ,为菌种的自动分类识别提供了理论依据和技术支持 ,也在计算机图像识别领域做出了新的尝试 .The microbial germ is very important in fermentation industry.It is always screened and optimized in order to reduce costs. The appearance of the colony is related to its biochemical performance, and is helpful to manual screening in industrial production.If the computer vision is applied instead of man eyes in germ classification and recognition, the efficiency will be improved obviously. In the paper a method for colony image classification is presented, taking green muscardine fungus for example. After preprocessing the source colony image, the middle part of the image was drawn out which is said to be the most representative.On the other hand, the original colony image was decomposed into some subimages by multiresolutional wavelet transforming, and these subimages include detailed information of the original one. The paper shows how to draw the texture entropy features from these subimages, and to select the best feature combination. Then, a linear classifier is designed. Through training and testing, the precision of the classifier is high more than 90%.

关 键 词：工业微生物 菌种 自动识别 小波变换 纹理熵 线性分类器 菌落显微图像 

分 类 号：TQ920.1[轻工技术与工程—发酵工程] Q939.97[生物学—微生物学]