Research on Skin Texture Classification by Gray Level Co-occurrence Matrix and the BP Neural Network

DOI

10.2991/tmcm-15.2015.8How to use a DOI?

Keywords

skin texture; gray level co-occurrence matrix; principal component analysis; BP neural network classifier

Abstract

It’s very common to use the skin texture of gray level co-occurrence matrix to calculate the four most representative eigenvalues of human facial skin image: energy, moment of inertia, correlation and entropy. To test whether the four eigenvalues can represent the skin texture information, the article designed a verification experiment: the article used comparison data included arithmetic average roughness(Ra), average roughness(Rz), and smooth depth data(Rt) measured from DERMATOP V3 of CK in Germany, and experimental data included the four eigenvalues, to do principal component analysis, respectively, for unrelated principal component as the input data of BP neural network classifier. The experimental results show that using the four eigenvalues, the classification accuracy is higher. The method using gray level co-occurrence matrix to extract facial skin texture eigenvalue can relatively reflect the degree of human facial texture state than texture information measured by DERMATOP V3, which provides a simple and effective method for the data acquisition of skin texture.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Qiaohua Liu
AU  - Tianhua Chen
AU  - Xiaoyi Wang
AU  - Jiping Xu
AU  - Li Wang
AU  - Yinmao Dong
AU  - Hong Meng
PY  - 2015/11
DA  - 2015/11
TI  - Research on Skin Texture Classification by Gray Level Co-occurrence Matrix and the BP Neural Network
BT  - Proceedings of the 2015 International Conference on Test, Measurement and Computational Methods
PB  - Atlantis Press
SP  - 26
EP  - 29
SN  - 2352-538X
UR  - https://doi.org/10.2991/tmcm-15.2015.8
DO  - 10.2991/tmcm-15.2015.8
ID  - Liu2015/11
ER  -