基于BP神经网络的脸部皮肤图像质量评价  

Facial Skin Image Quality Appraisal Based on BP Neural Network

作　　者：何煦佳[1] 张密[1] 吴效明[1] 

机构地区：[1]华南理工大学生物科学与工程学院,广州510006

出　　处：《医疗卫生装备》2010年第1期32-33,共2页Chinese Medical Equipment Journal

基　　金：广东省科技计划基金(2007B031302003)

摘　　要：目的:准确方便地进行脸部皮肤质量评价。方法:基于改进的BP神经网络算法,建立脸部皮肤图像质量评价模型,使模型不断学习训练样本中存在的内在模式,训练成功后即可通过各项皮肤指标的输入,输出评价结果。结果:采用23组样本进行实验,评价准确度达95%。结论:该网络模型有较高的评价精度,较低的误差率,具备实用价值。Objective To appraise facial skin＇s quality conveniently and correctly. Methods The quality appraisal model for facial skin image was established based on improved BP nerve network algorithm, and the model was trained with the intrinsic pattern of the training sample, then the appraisal result was output after inputting indexes on the skin. Results For 23 sets of samples, the appraisal accuracy reached 95%. Conclusion The network model, with a high accuracy and a low error rate, is applicable to facial skin image quality appraisal.

关 键 词：BP网络 皮肤图像 质量评价 非线性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]