首页|改进PCA方法的牙颌特征提取和数据集构建

改进PCA方法的牙颌特征提取和数据集构建

扫码查看
原文链接
  • 万方数据
  • 维普
为了提高牙颌数字模型特征提取的准确性和牙齿分割的效率,提出一种几何变换和主成分分析相结合的坐标系标准化方法,并以离散曲率、法向量、形状直径函数和离散测地距离为基础特征对牙颌数字模型进行了特征提取,在此基础上进一步扩展构建了 76 个特征数据集.采用提出的坐标系标准化方法和数据集对上颌进行了牙齿分割实验.结果表明:改进主成分分析方法能够快速准确地实现牙颌数字模型坐标系对齐,能够准确识别牙齿的特征信息并作出标记,牙齿分割完整,平均分割准确率达到 99.74%.基于改进主成分分析方法的牙颌模型特征提取方法能够极大地提高特征对牙齿的区分性,从而降低位姿对牙颌特征提取带来的负面影响,实现在特征数据较少的情况下准确分割牙齿,可为数字化口腔诊疗提供一定的参考.
Dental feature extraction and data set construction based on improved PCA method
To improve the accuracy of feature extraction of dental digital model and the efficiency of tooth segmentation,a coordinate system standardization method combining geometric transformation and principal component analysis was proposed.Based on discrete curvature,normal vector,shape diameter function and discrete geodesic distance,the feature extraction of dental digital model was carried out,and 76 feature data sets were further expanded and constructed.The proposed coordinate system standardization method and data set were used to perform tooth segmentation experiments on the maxilla.The results show that the improved principal component analysis method can quickly and accurately align the coordinate system of the dental digital model,accurately identify the feature information of the teeth and mark them.The teeth are segmented completely,and the average segmentation accuracy is 99.74%.The feature extraction method for dental model based on improved principal component analysis can greatly improve the discrimination of features to teeth,so as to reduce the negative impact of pose on dental feature extraction,and realize accurate segmentation of teeth with less feature data,which can provide some reference for digital oral diagnosis and treatment.

computer graphicsdental digital modelcoordinate system standardizationprincipal component analysisdiscrete curvatureshape diameter functiondiscrete geodesic distance

张永弟、王浩楠、王伟志、赵立松、杨光

展开 >

河北科技大学机械工程学院,河北石家庄 050018

计算机图形学 牙颌数字模型 坐标系标准化 主成分分析 离散曲率 形状直径函数 离散测地距离

石家庄市科学技术研究与发展计划

161460601A

2024

10.7535/hbgykj.2024yx03007

河北工业科技
河北科技大学

河北工业科技

CSTPCD
影响因子：0.694
ISSN：1008-1534
年,卷(期)：2024.41(3)