基于机器视觉下的皮肤老化分级研究

发布时间：2018-04-15 05:27

本文选题：图像处理 + 皮肤老化　；参考：《生物医学工程学杂志》2017年03期

【摘要】：皮肤老化是人体衰老进程中最明显的标志,对其进行定性或定量评价具有重要意义,并可广泛应用于人体衰老程度研究以及抗衰老措施功效评价等领域。针对传统人为皮肤老化分级的主观性,本文探索用自组织映射(SOM)神经网络实现对皮肤老化程度的自动分级。首先,采用便携式数码显微镜获取人体前臂腹侧皮肤图像,经图像处理分析,提取皮肤纹理参数:皮沟平均宽度和交点个数,用于表征皮肤纹理老化的变化情况;其次,将纹理参数值输入SOM神经网络,用于网络训练学习及分级。结果显示,本文所设计的基于机器视觉下的皮肤老化评价方法,与人工方法相比较,分类一致率达80.8%,可实现较为客观且快速的皮肤老化分级。
[Abstract]:Skin aging is the most obvious sign in the process of human aging. It is of great significance to evaluate it qualitatively or quantitatively, and it can be widely used in the research of human aging degree and the effectiveness evaluation of anti-aging measures.Aiming at the subjectivity of traditional artificial skin aging classification, this paper explores the use of self-organizing mapping (SOM) neural network to realize the automatic grading of skin aging degree.Firstly, the skin images of the ventral forearm of human body are obtained by portable digital microscope. The skin texture parameters, such as the average width and the number of intersection points, are extracted by image processing and analysis, which are used to characterize the aging of skin texture.The texture parameters are input into SOM neural network for network training and grading.The results show that compared with the artificial method, the classification consistency rate of the skin aging evaluation method designed in this paper is 80.8, which can achieve more objective and rapid skin aging classification.
【作者单位】：重庆医科大学医学信息学院;
【基金】：重庆市科委资助项目(cstc2013yykfA0095973)
【分类号】：R318;TP391.41

【相似文献】