Research And Implementation On The Recognition System Of Human Helminth Eggs Based On Hadoop

In recent years,the human parasitic diseases have been harmful to human health,and pathogen detection is the most common and important method for the diagnosis of parasitic disease.The traditional parasitic diseases detection method is mainly carried out by professionals to complete,which is not only cumbersome but also inefficient.With the rapid development of medical microscopic image processing technology and pattern recognition technology,researchers have proposed many automatic identification methods about helminth eggs based on computers,which have important significance to improve the working efficiency of the medical staff and inhibit the parasitic diseases.Although at present many scholars have done a lot of work in parasitic disease pathogen recognition,but in practical application some detected images hava a lot of impurities or complex background,which will bring great influence to image segmentation and feature extraction and affect the accuracy of the final recognition result.Therefore, this paper analise the related content in depth based on the previous research results to the helminth egg images, composites grayscale and color distribution of the image, and then combines a parallel computing model Map Reduce and puts forward a human helminth eggs recognition algorithm based on Hadoop which can not only improve the recognition correct rate,but also strengthen the system efficiency. The main work is as follows:(1) In practical application,helminth egg images always hava a lot of impurities and complex background,so this paper introduces the edge space distribution histogram matching mechanism to extract the edge of helminth eggs,and on this basis further extracts geometric feature,gray feature and the duty ratio,finally uses support vector machine as the recognition classifier and together the related features of gray image and chromatic image to propose a human helminth eggs recognition algorithm based on SVM.(2) In order to accurately detect the helminth eggs information in the picture,the recognition system needs to iterate through the edge space distribution histogram templates of all kinds helminth eggs,which will lead to the system identification efficiency declining with the increasement of the templates number.Therefore,this paper combines the helminth eggs recognition algorithm based on SVM with Map Reduce,and proposes a parallel human helminth eggs recognition algorithm based on Hadoop,which is effectively improve the identification efficiency of the system.(3) On the basis of the above work,we use several ordinary PC to build a Hadoop cluster,then design and implement a human helminth eggs recognition prototype system based on Hadoop,finally take an experimental analysis on 10 kinds common human helminth egg images,which is compare the distribued recognition system with the single recognition system.The experimental result shows that,this system can effectively improves the accuracy and efficiency of the helminth egg images recognition.