激光生物学报, 2018, 27 (1): 16, 网络出版: 2018-04-23  

人体皮肤的双光子图像纹理特征提取方法

Texture Feature Extraction of Two-photon Image of Human Skin
作者单位
医学光电科学与技术教育部重点实验室, 福建省光子技术重点实验室, 福建师范大学, 福建 福州 350007
摘要
近年来通过计算机技术对瘢痕实现无损诊断的研究进展迅速, 其对瘢痕图像纹理特征的量化分析得到了很好的诊断效果。在这个过程中出现了很多纹理描述方法, 这些方法的提出也促进了纹理研究的发展。本文在对灰度共生矩阵(GLCM)、局部三值模式(LTP)等统计纹理分析方法进行介绍的情况下, 在瘢痕图像上利用梯度迭代回归树算法给出了不同方法的实验结果, 得到了不同方法的回归模型。这些模型的性能体现在对不同年龄瘢痕的预测能力, 其中局部差异局部二值模式(LD-LBP)和局部方向三值模式(LOTP)得到的模型预测能力最好, 说明它们是目前对瘢痕图像纹理描述比较准确的方法之一, 同时表明统计纹理分析方法适合用于瘢痕图像的纹理研究。
Abstract
In recent years, the rapid progress of non-destructive diagnosis of scar has been achieved by computer technology, and the quantitative analysis of scar image texture has been well diagnosed. In this process there have been many texture description methods, the proposed method also contributed to the development of texture research. In this paper, we introduce the experimental results of different methods on the scar image by using the gradient iterative regression tree algorithm, and get the regression model of different methods in the case of gray level co-occurrence matrix(GLCM) and local ternary pattern(LTP) statistical analysis methods. The performance of these models is reflected in the ability to predict the scars of different ages. The local difference local binary pattern (LD-LBP) operator and the local orientation ternary pattern(LOTP) operator have the best predictors of the model, which shows that they are one of the most accurate methods for describing the texture of scar images. The statistical texture analysis method is suitable for the texture study of scar images.

刘高强, 李吉春, 林海洪, 余浩天, 陈冠楠. 人体皮肤的双光子图像纹理特征提取方法[J]. 激光生物学报, 2018, 27(1): 16. LIU Gaoqiang, LI Jichun, LIN Haihong, YU Haotian, CHEN Guannan. Texture Feature Extraction of Two-photon Image of Human Skin[J]. Acta Laser Biology Sinica, 2018, 27(1): 16.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!