赤潮藻类流式图像自动分析算法的研究  

Algorithm for Automatic Recognition of Red Tide Algal Images Captured by Flow Cytometry

作　　者：谢杰镇[1] 骆庭伟[1] 戴君伟[1] 王迪[2] 高艳 冉升 

机构地区：[1]厦门大学海洋与地球学院,厦门361005 [2]厦门大学信息科学与技术学院,厦门361005 [3]重庆西信天元数据资讯有限公司,重庆401127

出　　处：《计算机科学》2013年第7期293-296,共4页Computer Science

基　　金：国家863计划(2012AA092002)资助

摘　　要：赤潮是常见海洋自然灾害之一。为了早期预测和预报赤潮的发生,开发了基于流式图像的赤潮监测仪器,它将流式细胞、显微成像以及图像处理技术结合起来,采用基于背景差的方法快速准确地分割出藻类图像;为了克服藻类细胞处于不同生长周期和环境带来的形态和个体差异的影响,提取具有平移、旋转和尺度不变性的几何形状特征与基于灰度共生矩阵的纹理特征,采用一对一的多分类支持向量机进行分类识别。实验结果表明,该算法实现对了海洋原甲藻、红色裸甲藻、拟菱形藻和中肋骨条藻的自动分类,平均识别准确率高达94.37%。The red tide is a global marine natural disaster. In order to predict and forecast the occurrence of red tide, with the technology of flow cytometry, micro-image and image processing, a real-time harmful algae monitoring system was developed. A method based on background subtraction was used to quickly and accurately segment the algae ima- ges. In order to overcome the influence of algal cells in morphology and individual difference brought by the different growth period and environment, geometry features which are invariant to translation, rotation and scale and texture fea- tures based on GLCM were extracted. And finally, one-to-one multi-class support vector machine was adopted for iden- tification. The experimental results show that the average recognition accuracy rate is as high as 94. 37%.

关 键 词：赤潮 图像分割 特征提取 藻类识别 支持向量机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]