复杂背景中的人脸检测与定位

【作者】 吴艳冬；

【导师】 孙怡；

【作者基本信息】 大连理工大学 ， 通信与信息系统， 2002， 硕士

【摘要】 近些年来，自动人脸检测与识别在图像处理和计算机视觉领域引起许多学者的关注。这一课题在智能监视系统、虚拟现实、高级用户接口、表情分析和基于模型的图像编码等方面具有广阔的应用前景。利用图像或图像序列进行人脸检测与识别包含三个基本内容：(1)从背景中检测定位出人脸；(2)人脸跟踪；(3)人脸的识别和理解。其中，人脸的检测与定位是自动人脸检测与识别过程的关键，是进一步识别和理解人脸的基础。复杂背景中的人脸检测是自动人脸检测与定位的高级阶段，也是近几年来人脸检测研究的热点。本文实现了两种复杂背景中的人脸检测定位算法。 算法Ⅰ基于色彩信息，利用模糊模式匹配的方式检测人脸。该算法由两个部分组成。首先，粗略定出可能存在人脸的区域，即候选人脸区域；然后在候选人脸区域里细致检测以证实人脸的存在并定位证实存在的人脸。算法Ⅰ可检测彩色图像中任意背景，任意姿态(-90°—90°)，多尺度，不同亮度条件下的任意个人脸。 算法Ⅱ基于支持向量机(SVM)，由训练和检测两部分组成。训练过程是用大量人脸样本、“非人脸”样本训练SVM分类器，使之获得一个最优分类超平面。检测阶段用训练好的SVM分类器检测图像中的人脸。算法Ⅱ可检测彩色或灰度图像中任意背景，大致正面姿态，任意表情，多尺度的任意个人脸。 论文分别给出了使用算法Ⅰ、算法Ⅱ检测人脸的实验结果，并对算法Ⅰ与算法Ⅱ进行了检测性能的比较。实验结果证明算法Ⅰ的检测性能优于算法Ⅱ。更多还原

【Abstract】 Automatic face detection and recognition has been receiving increasing attention from researchers in the fields of image processing and computer vision during the past few years. It has a lot of applications in smart surveillance system, virtual reality, advanced user interface, emotion analysis and model-based coding, etc. Generally speaking, the procedure of the automatic face detection and recognition from images or a sequence of images involves three main stages: (1) face detection and location in a scene; (2) face tracking; (3) face recognition and understanding. As the base of the face detection and recognition, face detection and location is the key of the whole procedure. Face detection in complex backgrounds is the advanced stage of the automatic face detection and location. And it is focus in the recent face detection research. This paper describes two methods of the face detection in complex backgrounds automatically.Algorithm I is based on color information, and constitutes two sections. First, the technique approximately detects the image positions where the probability of finding a face is high; then the location accuracy of the candidate faces is improved and their existence is verified and marked. It can be used to detect many faces with differentsizes and directions ( - 90 - 90) in a color static image.Algorithm II is based on support vector machine and made of training section and detecting section. A lot of face samples and "not face" samples are used to train the SVM classifier, to get optimal separating hyperplane in the training. And SVM classifier is used to detect faces in the detecting. This algorithm can be used to detect many frontal faces with different sizes in a color image or gray image.The paper respectively gives the experimental results of algorithm I and algorithm II, and compares the detection performance of the two algorithms. Experimental results show that algorithm I is better than algorithm II.更多还原