改进的统计模型三维人脸特征点标定算法框架

发布时间：2018-09-17 14:58

【摘要】：自动三维人脸特征点标定是计算机视觉领域的研究热点,其广泛应用于人脸识别,人脸模型配准,表情识别,脸部动画等领域。通过对三维人脸样本统计建模,采用遗传算法对待匹配模型的生成数目进行参数优化,利用模型相似性匹配方法及其映射关系对三维人脸特征点进行自动标定。首先,对三维人脸数据预处理,然后对其统计建模并通过模型形变得到有映射关系的基准模型和待匹配模型。利用遗传算法对待匹配模型中的待匹配模型生成数目参数进行优化,生成与之对应的待匹配模型数;接着计算待测模型与待匹配模型的相似度。最后,利用模型相似度和模型映射关系,间接得到待测模型的特征点。实验结果表明,提出的算法是可行的,能够在一定程度上提高原有算法的效率。该算法可以自动标定三维人脸模型的特征点,当距离阈值为10像素时,39个三维人脸特征点定位的准确率都可以达到100%,并有效解决了传统方法中三维人脸模型平滑区域特征点精度不高的问题。
[Abstract]:Automatic 3D face feature point calibration is a hot topic in the field of computer vision. It is widely used in face recognition, face model registration, facial expression recognition, facial animation and so on. Through the statistical modeling of 3D face samples, the genetic algorithm is used to optimize the parameters of the generated number of matching models, and the model similarity matching method and its mapping relationship are used to calibrate the 3D face feature points automatically. Firstly, the 3D face data is preprocessed, and then the statistical model is modeled and the corresponding datum model and the matching model are obtained by the deformation of the model. Genetic algorithm is used to optimize the number of models generated in the matching model to generate the corresponding number of matching models, and then calculate the similarity between the model to be tested and the model to be matched. Finally, the feature points of the model under test are obtained indirectly by using model similarity and model mapping relationship. Experimental results show that the proposed algorithm is feasible and can improve the efficiency of the original algorithm to some extent. The algorithm can automatically calibrate the feature points of 3D face model. When the distance threshold is 10 pixels, the accuracy of all the 39 3D facial feature points can reach 100%, which effectively solves the problem that the accuracy of feature points in the smooth region of 3D face model is not high in the traditional method.
【作者单位】：荆楚理工学院计算机工程学院;中国矿业大学银川学院机电动力与信息工程系;
【分类号】：TP391.41

【相似文献】