一种由粗到细的渐进式特征点定位算法

    A Kind of Accuracy Increasing Lip Feature Point Location Algorithm

    • 摘要: 为了解决描述口形轮廓的特征点定位问题,提出了一种特征点定位算法,通过建立口形灰度图像与口形特征点矢量之间的线性映射关系,实现了特征点的初始定位.在此基础上利用局部纹理模型,实现了对特征点位置的进一步准确调整.该算法克服了单纯基于局部纹理模型可能陷入搜索过程中局部收敛的缺点并提出了适合描述唇部特征的局部纹理模型,提高了特征点标定的准确性.

       

      Abstract: In order to solve the feature points location problem, which are used to describe the month appearance, a kind of feature location algorithm was proposed. It realizes the preliminary feature point location through establishing the affine relationship between lip gray scale image and lip feature point vector. Based on it, the local texture model is used to adjust every point s accurate location. The feature point location algorithm the paper proposed overcomes the shortage of traditional local texture model based only algorithm, which may get in the local convergence. The paper proposes a kind of local texture model which is suitable to describe the lip appearance and improves the accuracy of feature point location.

       

    /

    返回文章
    返回