基于BP神经网络模型的临床瓷粉配色方法

发布时间：2018-03-09 23:23

本文选题：BP神经网络　切入点：瓷修复体　出处：《武汉大学学报(医学版)》2014年03期 　论文类型：期刊论文

【摘要】：目的:探讨基于BP神经网络预测瓷粉配方的配色网络模型的可行性。方法:按照不同质量配比的混合松风Halo瓷粉制作32个瓷片样本,随机分为两组:训练组和测试组。用Matlab软件构建BP神经网络模型,用训练组瓷片训练并优化网络,形成颜色L*a*b*参数与瓷粉配方质量配比之间的非线性映射关系。用测试组试样检验构建的神经网络的精度。结果:测试组预测出的混合瓷粉质量配比与真实质量配比之间的准确率为80%(配比绝对误差±0.05)。按照测试组的预测和真实瓷粉质量配比制作的瓷片之间的平均色差为1.68(远小于临床可容忍阈值2.7)。结论:基于BP神经网络模型的瓷粉配色方法,直接给出了瓷粉配方,缩减了操作程序,减小色差,为当前口腔临床修复技术提供了一种新的比色途径。
[Abstract]:Objective: to explore the feasibility of the model of color matching network based on BP neural network to predict the formula of porcelain powder. The BP neural network model was constructed with Matlab software, and the training group was trained and optimized by porcelain chip. The nonlinear mapping relationship between the parameters of the color LZAZB * and the quality ratio of the ceramic powder formula was formed. The accuracy of the neural network was tested with the test group. Results: the mass ratio of the mixed ceramic powder predicted by the test group and the real quality match were obtained from the test group. The accuracy of the ratio is 80% (absolute ratio error 卤0.05). The average color difference between the test group and the quality ratio of real porcelain powder is 1.68 (much less than the clinical tolerable threshold 2.70.Conclusion: based on BP neural network model, the average color difference between the two groups is less than that of the clinical tolerance threshold of 2.70.Conclusion: based on the BP neural network model. The method of matching the color of porcelain powder based on. The formula of porcelain powder is given directly, the operation procedure is reduced, and the color difference is reduced, which provides a new way of colorimetric for the current clinical prosthodontic technique.
【作者单位】：武汉大学口腔医院修复科;
【分类号】：R783.3

【参考文献】