基于线性判别分析的寄生虫卵识别

寄生虫病是危害人类及动物健康的疾病之一。为了实现对寄生虫卵的自动识别,辅助临床检测,提出基于线性判别分析的寄生虫卵识别方法。采用结合形态学滤波和Otsu的方法分割得到寄生虫卵及其轮廓,提取形状特征和纹理特征作为特征向量集,并利用线性判别分析实现对寄生虫卵自动识别。实验结果表明,该方法对6种寄生虫卵的识别正确率达到90.70%。

Parasite Eggs Recognition Based on Linear Discriminant Analysis

Parasite disease is a kind of disease that will do great harm to human and animal health. For the purpose of identification of parasite eggs to help clinical testing, presents an automatic parasite eggs recognition based on linear discriminant analysis. Uses morphological filtering and Otsu＇s method to segment the parasite eggs and extract its contour, extracts shape and texture features to construct the feature vector set, and performs linear discriminant analysis for automatic identification of parasite eggs. Experimental results show that average recognition rate of 7 species of parasite eggs can reach 90.70%.