基于特征值修正的特征点重定位方法  

Feature points re-localization method based on eigenvalue correction

作　　者：刘小琴[1] 赵晖[1] 

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《计算机工程与设计》2013年第6期2051-2056,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(60962005;61261037)

摘　　要：为了解决特征点重定位算法的运行时间过长的问题,研究了人脸活动单元(action units,AU)的动态性,提出一种通过修正特征值来对特征点进行重定位的方法,从数学角度对特征值进行操作。该方法在主动表观模型(active appearance model,AAM)自动定位之后,根据整个序列中某一活动单元特征值的变化趋势来对序列进行去孤立帧、平滑等操作,进而通过计算间接地对特征点进行重新定位。实验结果表明,该方法避免了直接调整特征点的操作复杂性,在一定程度上达到提高特征点定位精确度的目的。To solve the problem that relocating feature points take for quite a long running time, the dynamic for facial action units is studied and a method on relocating feature points through a modified characteristic value is presented. Eigenvalues are operated from a mathematical point of view. After automatic positioning of the AAM, according to the entire change trend of the eigenvalue of an action unit, isolated frames are removed and the sequence is smoothed. The experiment shows that the method not only reduces the complexity of directly adjusting the feature point, but also achieves the purpose in a certain extent to improve the positioning accuracy of the feature point.

关 键 词：活动单元 主动表观模型 特征值 重定位 孤立帧 平滑 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]