Abstract: human Face detection and segmentation is an important process in the system of face analysis and recognition. It’s purpose is to detect and segment human face from an image with complex background. It is easy for human to accomplish this work, but it is very difficult for an automatic system. In the past years, many human face detection methods were studied with the progress of computer science and the methods of image processing. This paper applied a method of human face detection based on YIQ color model and a method of pose adjustment with the inertia minimum theory in order to segment the face area preferably. Experiments showed that these methods can detect human face from complex background and can segment it efficiently.

Key words: human face detection, YIQ color model, pose adjustment, inertia minimum theory

摘要： 人脸检测和分割是进行人脸分析和识别的重要组成部分，其目的是从复杂背景图像中检测出人脸的位置，并把人脸分割出来。虽然人们可以毫不困难地在复杂背景中检测和识别出人脸，但是要建立一套完全自动化的人脸和识别系统却是非常困难的。近年来随着计算机技术和图像处理方法的不断发展，出现了各种不同的人脸检测方法。本文应用了一种基于YIQ颜色模型的人脸检测方法，并且利用惯量最小化原理进行人脸图像的姿态调整，然后进行分割提取。试验结果表明，该方法能较好地从复杂背景中检测出人脸区域并能较完整地分割出人脸。

关键词: 人脸检测, YIQ肤色模型, 姿态调整, 惯量最小化原理