基于特征加权支持向量机的血吸虫尾蚴识别算法设计与应用  

Design and Application of Schistosome Cercariae Recognition Algorithm Based on Feature Weighted SVM

作　　者：易云[1] 汪廷华[1] 

机构地区：[1]赣南师范学院数学与计算机科学学院,江西赣州341000

出　　处：《赣南师范学院学报》2012年第3期37-40,共4页Journal of Gannan Teachers' College(Social Science(2))

基　　金：国家科技部科技支撑计划基金(2009BAI78B01)

摘　　要：本文提出一种利用特征加权支持向量机(FWSVM)识别河流、湖泊等水域中血吸虫尾蚴的算法.该算法采用图像形态学、动态阈值化等图像处理方法在图片中搜索可能存在尾蚴的区域,然后利用疑似区域图像的Hu矩作为特征值训练FWSVM,最后利用训练好的分类器对采集的视频进行识别.实验结果表明该算法能够有效识别水域中的血吸虫尾蚴.This paper proposed an algorithm that using feature weighted support vector machine（FWSVM） to automatically distin- guish the schistosome eercariae in rivers and lakes. After had gotten the video under the microscope, we used approaches such as image morphology and dynamic threshold to search image regions where there are cereariae, and then used the Hu moments of the image as the feature to train FWSVM, finally used the classifer which had been trained to identify the video that had been collected. Experimen- tal results show that the algorithm can successfully identify the schistosome cercariae in water.

关 键 词：血吸虫尾蚴 模式识别 HU矩 特征加权 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]