寄生虫虫卵图像计算机自动识别技术研究

目的探讨建立寄生虫虫卵显微图像计算机自动识别系统的相关问题。方法从研究国内外虫卵图像识别技术的相关进展和10种人体常见寄生虫虫卵图像的特性入手,以Visual C++6.0为系统开发工具,运用数字图像处理,将虫卵数字图像经过预处理,提取形态、颜色和纹理等特征,通过相应的虫卵分类识别算法的实现,初步实现人体常见寄生虫虫卵图像的计算机自动识别。结果实验系统可在获取图像后一步实现对10种常见人体寄生虫虫卵的自动识别,平均识别率为93.0%,每幅图像处理时间为1~3s。结论对虫卵图像纹理特征的选用,可显著提高系统对样本的分类精度;系统开发所采用的分析方法、特征和计算参数的选取原则,有益于日后的类似研究。

Engineering Research on the Automatic Identification of Helminth Egg Images

Objective To explore how to develop a micro-image recognition system for helminth eggs.Methods The relevant progress about the automatic image recognition technology of helminth eggs and the image's features of 10 common human helminth eggs were researched.Taking Visual C++6.0 as the system development kit,using digital image processing technology and passing the pretreatment,some features of 10 common human helminth eggs' digital images were extracted,such as conformation,color,texture,etc.Through corrsponding categorized algorithm to helminth eggs in the computer,the automatic image recognition of human common helminth eggs was developed tentatively.Results The experimental system can realize the automatic recognition to 10 common human helminth eggs in one step after image gain,the average recognition rate of accuracy is 93.0%,the process time is 1～3 seconds per image.Conclusion The exertion of texture features of eggs'image can raise the categorized precision to the sample notably in the system.The analytical method about human body helminth eggs'micro-image,the selecting rinciple about the image features and calculation parameter proposed in this thesis would be beneficial to similar research in the future.