基于手机终端的舌诊色差校正函数模型研究

Study on the functional model of color calibration of tongue diagnosis based on mobile phone terminal

目的:探讨建立基于手机终端的舌象图像色差校正函数模型。方法:本研究以TDA-1型小型舌诊仪为参照标准,通过5种不同型号的手机设备在实验最优条件下,对349例舌象进行标准色卡L^(*)a^(*)b^(*)色值采集与分析,建立并优化反向传播(BP)神经网络模型。结果:分析舌诊仪与不同手机拍摄标准色卡所得平均色差值,筛选P牌手机进行临床采样。通过BP神经网络校正,舌质、舌苔色值与标准值差值显著降低(P<0.05),校正后ΔE均显著减小(P<0.01),颜色分界线清晰。结论:此次基于手机终端建立的舌诊色差校正函数模型所得预测输出值接近由TDA-1型小型舌象仪拍摄所得L^(*)a^(*)b^(*)值,因此该校正模型可有效进行色差校正以提高手机所拍摄舌象图像的色值精确度。

Objective:To explore the establishment of a function model for color calibration based on the tongue image of a mobile phone terminal.Methods:In this study,the TDA-1 small tongue imager was used as a reference standard,and the standard color card L^(*)a^(*)b^(*)color value collection and analysis were carried out through 5 different types of mobile phone devices under the optimal conditions of the experiment.Then we built and optimized the BP neural network model.Results:Analyze the average color difference value obtained by the tongue diagnostic instrument and different mobile phones for shooting standard color cards,and screen the P brand mobile phones for clinical sampling.Corrected by BP neural network,the difference between tongue quality,tongue fur color value and standard value was significantly reduced(P<0.05).After correction,ΔE was significantly reduced(P<0.01),and the color boundary was clear.Conclusion:The predicted output values of this model are close to the L^(*)a^(*)b^(*)values obtained by the TDA-1 small tongue imager.So the correction model can effectively perform color difference correction to improve the accuracy of the color value about the tongue image based on mobile phone.