摘要
建立基于MATLAB的牙齿比色分类算法,提高牙齿修复体与天然牙颜色的匹配度;在自然光下随机平面拍摄来获取样本,建立牙齿数据库,通过投影法分割出固定的牙齿中心区域,取红绿蓝、色调-饱和度-明度、亮度-颜色通道三种色彩空间各分量作为特征值,然后采用光照自适应校正算法和方差选择特征方法,得到最优分类特征组,进行K最近邻分类;光照自适应校正算法有效减少了光照的影响,提高了分类准确率,通过方差选择特征法和KNN分类验证,得到本实验中分类最优特征组由蓝-色调-明度三个特征值组成,该特征组达到最高的分类准确率96.2%;基于MATLAB牙齿比色分类算法实现了较高的分类准确率,为实现牙齿智能比色系统提供理论依据。
To establish a set of dental colorimetric classification algorithms based on MATLAB to improve the matching degree of dental restorations and natural tooth colours;Obtain samples by random plane shooting under natural light,establish a tooth image database,and cut up the fixed central area of the tooth by projection method. Take each component of the three colour spaces of red-green-blue,hue-saturation-value,and CIELab as eigenvalues. The adaptive illumination correction algorithm and the variance selection feature method are used secondly to obtain the optimal classification feature groups by K-NearestNeighbor classification. The adaptive illumination correction algorithm effectively reduces the interference of illumination,and the classification accuracy rate is improved. Through the variance selection feature method and KNN classification verification,it is manifested that the optimal feature set for classification in this experiment is composed of three feature values of Blue-Hue-Value. This feature group achieved the highest classification accuracy of 96.2%;Based on MATLAB,the dental colorimetric classification algorithm established in this paper achieves a higher classification accuracy rate,thus providing theoretical basis for future realization of intelligent tooth colorimetric system.
作者
杨世波
应梦迪
夏秋婷
李宏
YANG Shibo;YING Mengdi;XIA Qiuting;LI Hong(College of automation,Hangzhou Dianzi University,Hangzhou Zhejiang 310018,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2021年第5期629-634,共6页
Chinese Journal of Sensors and Actuators
关键词
牙齿比色分类
光照自适应校正算法
方差选择特征法
KNN
MATLAB分析
dental colorimetric classification
adaptive illumination correction algorithm
variance selection feature method
KNN
MATLAB analysis