节点文献
人脸定位研究与应用
Research and Application on Face Detection
【作者】 徐健；
【导师】 寇卫东；
【作者基本信息】 西安电子科技大学 ， 通信与信息系统， 2006， 硕士
【摘要】 人脸定位是人脸识别的第一步。定位的好坏直接决定着后续识别的准确程度。由于人脸的角度,遮蔽和阴影等非线性因素导致了多角度人脸识别比固定角度(例如正面)的人脸识别困难得多。本文就两种人脸识别方法分别利用不同的数学工具设计了两种不同的分类器。针对正面人脸识别的分类器使用两步定位法,先进行粗定位再进行细定位。粗定位算法的主要步骤是,先对图像进行预处理,再使用OAC滤波器滤波,最后使用阈值分割法将图像分割为人脸区域、可能人脸区域和非人脸区域三部分。为人脸细定位算法减少了计算量,增加了准确率。对CMU数据库中的图像进行测试,非人脸区域的排除率达到99%,误检率仅有1.3%。实验结果证明了该算法的有效性和可靠性。针对多角度人脸识别,角度空间被分为多个部分。每部分设计一种分类器。无论是否是人脸,都要使用支持向量回归进行人脸角度估计。人脸角度估计的结果用于分类器来估计是否是人脸。由于姿势估计这一步,大大减少了计算量。同时,由于每个分类器都只针对一部分角度的人脸,所以提高了准确率。经过对比,我所用的算法同现存人脸定位方法中速度较快的特征脸法相比,具有同样的速度,而且具有更高的准确率。
【Abstract】 Detection is the first step of face recognition, which directly affects the result of face recognition. Detecting faces across multiple views is more challenging than in a fixed view, e.g. frontal view, owing to the significant non-linear variation caused by rotation in depth, self-occlusion and self-shadowing.In this paper, I apply two classifier using two different methods, one for frontal view detection and the other for multiple views.The first method uses two-step classifier. The first step is rough detection, and the second step is delicate detection. The main processes of the rough detection algorithm are that, firstly, preprocess the image, secondly, filter in Fourier transform domain by OAC, thirdly, segment the image into "face" area, "may-be face" area and "none-face area". The testing on CMU face database shows that the algorithm successfully segments the image with "non-face" rejecting rate of 99% and false alarm rate of 1.3%. The experiment results confirm that the algorithms are efficient and reliability.For the second method the view sphere is separated into several small segments. On each segment, a face detector is constructed. We explicitly estimate the pose of an image regardless of whether or not it being a face. A pose estimator is constructed using Support Vector Regression. The pose information is used to choose the appropriate face detector to determine if it is a face. With this pose-estimation based method, considerable computational efficiency is achieved. Meanwhile, the detection accuracy is also improved since each detector is constructed on a small range of views. We developed a novel algorithm for face detection by combining the Eigenface and SVM methods which performs almost as fast as the Eigenface method but with a significant improved speed.