视频图像中人脸姿态估计的快速算法  

A Fast Algorithm for Face Pose Estimation in Video Image

作　　者：俞骋 YU Cheng(Ningbo No.2 Technical College, Ningbo 315012)

机构地区：[1]宁波第二技师学院,宁波315012

出　　处：《现代计算机》2018年第8期46-48,53,共4页Modern Computer

摘　　要：提出一种人脸姿态估计快速算法。该算法首先用肤色分割确定人脸和嘴部的位置,然后通过嘴部在正视图上的投影变化估算出人脸的姿态。与当前主流的人脸姿态估计算法相比,该算法只需经过简单的数学计算,时间和空间复杂度小,且具有较高的鲁棒性,非常适合应用于视频图像的检测。Proposes a new algorithm which can quickly estimate human＇s face pose. The algorithm detects the face and mouth in image with color segmentation algorithm firstly, and then estimates the pose of face according to the changes of mouth＇s projection in orthographic view. Compared with the mainstream face pose estimation approaches currently, the proposed algorithm only relies on simple mathematical operations, which makes it to be relative low time and space complexity. Outcomes of the tests also indicate that this algorithm is robust. All the merits make this algorithm to be suitable for the detection in video image.

关 键 词：人脸姿态估计 快速算法 视频图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]