欢迎访问《日用化学工业(中英文)》,今天是

日用化学工业 ›› 2020, Vol. 50 ›› Issue (7): 457-463.doi: 10.3969/j.issn.1001-1803.2020.07.005

• 开发与应用 • 上一篇    下一篇

PCA算法在面部微生态研究中的应用

高学义1(),王瑜1(),何聪芬2,冯春波3,陈圆圆3,宋丽雅2   

  1. 1. 北京工商大学 计算机与信息工程学院,北京 100048
    2. 北京工商大学 理学院,北京 100048
    3. 上海家化联合股份有限公司,上海 200438
  • 收稿日期:2019-10-23 修回日期:2020-06-04 出版日期:2020-07-22 发布日期:2020-07-23
  • 通讯作者: 王瑜
  • 作者简介:高学义(1995-),男,黑龙江海伦人,硕士研究生,电话:18811021319,E-mail: xueyi_g@163.com
  • 基金资助:
    国家自然科学基金面上项目资助(61671028)

Application of PCA algorithm in the research of facial micro-ecology

GAO Xue-yi1(),WANG Yu1(),HE Cong-fen2,FENG Chun-bo3,CHEN Yuan-yuan3,SONG Li-ya2   

  1. 1. School of Computer and Information Engineering, Beijing Technology and Business University, Beijing 100048, China
    2. School of Science, Beijing Technology and Business University, Beijing 100048, China
    3. Shanghai Jahwa Corporation, Shanghai 200438, China
  • Received:2019-10-23 Revised:2020-06-04 Online:2020-07-22 Published:2020-07-23
  • Contact: Yu WANG

摘要:

利用化学试剂法研究面膜对面部微生态的影响,存在过程繁琐、微生物种类不易确定等缺点,针对此问题,提出一种基于主成分分析(principal component analysis,PCA)的微生态分析方法。首先利用主成分分析对面部微生态数据进行降维处理,累计贡献率选取阈值为95%,确定变换空间下的特征向量以及特征维度k,然后根据特征向量与面部微生态数据的映射关系,此处阈值同样选取95%,确定面部微生态数据中对皮肤状态影响贡献率较大的属性,即改善面部皮肤状态的微生物种类。实验结果表明,该方法有效地克服了传统化学试剂法存在的弊端,能够快速、准确地确定面膜改善皮肤状态的微生物种类,同时也能挖掘不同面膜引起皮肤变化的差异,可为化妆品行业制造对面部皮肤更有益的面膜提供建设性意见。

关键词: 主成分分析, 降维, 特征选择, 面膜, 微生态

Abstract:

Some disadvantages such as cumbersome process and difficult microorganism confirmation exist when traditional chemical reagent method is used to study the influence of mask on facial micro-ecology. In order to solve this problem, a method of micro-ecological analysis based on principal component analysis (PCA) was proposed. First, PCA was used to reduce the dimensionality of the facial micro-ecological data which could determine the eigenvector and the dimension k when the cumulative contribution rate was 95%. Then, when the threshold value was also 95%, the mapping relationship between eigenvector and the facial micro-ecological data was used to determine the attributes with large contribution rate in the facial micro-ecological data, namely, the types of microorganisms which could improve the facial skin status. The experimental results showed that the proposed method could effectively overcome the shortcomings of the traditional chemical reagent method, and could be used to quickly and accurately determine the microorganism species which were related to the improvement of the skin status after using the mask. Meanwhile, the differences of the changes of microorganism species caused by different masks were also found, which could provide suggestions for the cosmetics industry to make more beneficial masks for facial skin.

Key words: principal component analysis, dimension reduction, feature selection, mask, micro-ecology

中图分类号: 

  • TQ658