朴素贝叶斯算法在女童体型判别中的应用

摘要

In order to achieve accurate identification of girl's figure,a large number of girl samples are collected to establish a database with detailed measurement index information.The data were analyzed and the girl's figure were divided into three types using cluster analysis method.On the basis,the Naive Bayesian (NB) algorithm was applied to the study of girl's figure discrimination,and a discriminant model was established.The discrimination of girl's figure was realized.At the same time,the similar body of the test sample in the database was found by utilizing minimal difference algorithm.And finally,the radar map was charted to fit the main measurement size of the both,using the main measurement size as the basic index for inquiry.The results show that the discriminant accuracy of girl's figure discriminant model established by the NB algorithm reaches 93.8％.And the error of the main measurement size of the test sample and the similar body is small in the radar map.The test sample can be replaced by the similar body,and the method provides data support for the relevant application areas needing to acquire detailed measurement information about girls.%为实现女童体型的准确判别归类,通过采集女童样本数据,建立了具有详细测量信息的数据库.利用聚类分析方法将女童体型分为3类,在此基础上,应用朴素贝叶斯算法建立了判别模型,实现了女童体型的判别归类.同时利用最小差值算法查找了被测女童样本在数据库中的相似体,最后,以主要控制部位尺寸作为查询的基础指标,绘制了雷达图拟合二者的主要控制部位尺寸.研究得出:朴素贝叶斯女童体型判别模型的判别准确度达93.8％:被测女童样本及其相似体在雷达图中各主要控制部位尺寸误差较小,相似体可用于替代被测样本;为需要获取人体详细测量信息的相关应用领域提供数据支持.