基于肤色相似度的人脸检测与定位-毕业论文设计.docVIP

基于肤色相似度的人脸检测与定位 摘 要 本课题致力于完成Visual C++ 6.0平台下的人脸检测与定位系统，人脸检测定位的算法限制于基于肤色相似度算法。要完成的工作包括相似度的计算，图像的二值化，垂直直方图和水平直方图的获取，人脸特征（眼睛、嘴和鼻子）的提取。本课题着重于肤色模型相似度在人脸检测与定位中的应用，对增加人脸检测与定位的准确率的研究有一定的指导意义。 提出一种基于肤色的人脸检测定位算法，设计了基于肤色的人脸检测和定位系统。 采用了脸部皮肤之间相似度的方法以及二值化方法，使用了基于边界方法和基于区域方法相结合的算法， 提取了眼睛、嘴和鼻子等关键特征，最终较好地实现了人脸定位。 在Microsoft Windows平台上，利用Visual C+ + 6. 0 开发了软件。 本课题的成果具有一定的应用价值。实验结果表明，该软件对于一定尺寸范围内清晰的正面人脸图能够正确检测定位并提取特征，并且在速度和准确性方面具有良好的性能。 关键词：图像分割；人脸定位；肤色；人脸检测；特征提取 Detection and Localization of Person Face Based on Skin Color Similarity Abstract This topic devotes to completing the detection and localization system of the person face under the Visual C++ 6.0 platforms, and the detection localization algorithm of the person face limit to basing on the skin color similarity algorithm. The topic must complete similarity computation, binary image processing, vertical histogram, horizontal histogram and extracting person face characteristic (eye, mouth and nose). This topic emphasize application of skin color model similarity in the detection and the localization of person face, and has a significant instruction for research of increasing accuracy in detection and the localization of person face. In this paper, the authors have presented an algorithm and designed a system for face detection and location based on complexion. By strengthening the contrast between face features and by adopting binary image processing method, the system has improved the preprocessing effect; and by using boundary-based algorithm plus region-based algorithm , the system has realized face location through the extraction of the features of eyes, nose and mouth. Taking advantages of Visual C++ 6.0, the authors have also developed corresponding software based on Microsoft Windows. Production of this paper have definite application value.Experiment results prove that the system is valid in detecting, locating and extracting frontal view face features in a certain range.And it possess