基于特征提取的人脸检测与定位技术.docx

摘要 人脸检测是指在图像或视频中判断人脸是否存在，若存在，确定人脸的大小、位置。人脸检测是实时人脸识别和表情识别的基础，只有将人脸检测准确，人脸识别和表情识别才得以实现。人脸检测还在视频监控、数字视频处理和基于内容的人脸检索中有着重要和广泛的应用。 计算机人脸识别技术也就是利用计算机提取人脸的特征来分析人脸图像，进而从中提取出有效识别信息，并根据这些特点进行人脸身份验证的技术。人脸与人体的其他生物特征（指纹、虹膜等）一样与生俱来，它们所具有的不易被复制和唯一性的良好特性为身份鉴别提供了必要的前提。 然而人脸作为人类自身在熟悉不过的一个组成部分，其检测问题却是一个极赋挑战性的课题。首先人脸是一个包含五官、毛发等的极不规则的复杂待测目标，不同的人脸在形状、大小、颜色、质地等方面都有很大的变化性；其次为了实现检测方法的鲁棒性，我们要考虑人脸在各种复杂的背景中，不同的方向、角度、尺度等情况下所展现出来的不同表象。 可见人脸检测有一定的难度和复杂性，但对这一问题的深入研究必将推动模式识别、计算机视觉、人工智能等计算机科学分支的发展。本文介绍了一种基于特征提取技术的人脸检测与定位系统，可分为人脸检测与定位两个模块。主要实现人脸的定位，人脸的眼睛、鼻子、嘴巴特征点的定位。运用了基于肤色的人脸检测算法，人脸区域分割算法，特征点的轮廓提取。 关键词：图像处理、人脸识别、特征提取、阈值分割 Abstract Face detection is an image or video face in judging whether there, if it exists, determine the face size, position. Face detection is the basis of real-time face recognition and expression recognition, only to face detection, face recognition and expression recognition was achieved. Face detection in video surveillance, digital video processing and content-based face retrieval has an important and wide application. Computer face recognition technology is the use of the characteristics of computer extracting face to face image analysis, and then extract the effective recognition information from, and for face authentication according to the characteristics of technology. Other biological features of face and body (fingerprint, iris etc.) as innate, they are not easy to be copied and uniqueness of good characteristics for authentication provides the necessary prerequisite. However the face as human beings in the familiar one component, the detection problem is a very challenging task. The first face is a facial features, hair and other irregular and complicated target, different face has changed greatly in shape, size, color, texture and so on; secondly, in order to achieve robustness detection method, we should consider the face in complex background, different representation in different direction, angle, scale and other circumstances revealed. Visible face detection has a