基于非接触式漫反射光谱的人体肤色检测研究

Research on Human Skin Color Detection Based on Non-Contact Diffuse Reflectance Spectroscopy

摘要: 当前有许多肤色检测技术出现，但在环境光变化的情况下很难检测出肤色。为了准确的检测肤色，本文提出了一种基于非接触式漫反射光谱的肤色检测方法，利用非接触式测量装置采集人体皮肤的漫反射光谱数据并将其作为测试样本，以采集的标准色卡的皮肤色号作为参考样本，随后建立支持向量机分类模型训练测试样本和参考样本，为测试样本定标色号，最后模型定标准确率达到了93.5%。结果表明，基于非接触式漫反射光谱的人体肤色检测，能够极大地提高肤色检测的准确性。

Abstract: At present, there are many skin color detection technologies, but it is difficult to detect skin color when the ambient light changes. In order to accurately detect skin color, a method of skin color detection based on non-contact diffuse reflection spectrum was proposed. Using a non-contact measuring device to collect the diffuse reflection spectrum data of human skin and use it as a test sample, with the color number of the standard color card collected as the reference sample, then support vector machine classification model was established to train test samples and reference samples, and the color number of test samples was calibrated. Finally, the accuracy of model calibration reached 93.5%. The results show that the detection of human skin color based on non-contact diffuse reflection spectrum can greatly improve the accuracy of skin color detection.