摘要
本文提出一种基于Shape Index特征和几何约束的三维人脸标志点定位方法。首先通过曲面拟合计算模型的连续Shape Index特征,对Shape Index特征进行阈值分割和区域标记,然后由统计方法建立的标志点几何约束模型进行区域筛选,最终得到标志区域及标志点。实验结果表明本文所提方法的平均定位准确度高于90%,具有较高的姿态鲁棒性。
A method of markers localization of 3D face based on Shape Index features and geometric constraints was proposed in this paper. Firstly, the continuous Shape Index features of model were computed after surface fitting, segmentation and region label was done with these Shape Index features, then the regions were selected by geometry constrained model, finally, the markers were obtained. Experiment shows our method is robust and its average accuracy higher than 90%.
出处
《微计算机信息》
2009年第7期264-266,共3页
Control & Automation
基金
基金申请人:毕笃彦
项目名称:高效图像数据压缩技术
基金资助项目:国家863计划(2007AA701206)