基于双肤色模型和AdaBoost的人脸检测系统设计

Design of Face Detection System Based on Dual Skin Models and AdaBoost Algorithm

本文在Windows平台下,基于VS2010和Intel开源计算机视觉库OpenCV设计了一个实用的人脸检测系统,实现复杂背景图像中可能存在的人脸区域检测。人脸检测功能的实现主要融合双肤色模型和AdaBoost算法。首先基于YCbCr颜色空间的简单边界模型和高斯模型对图像实现分割并分析处理,实现人脸区域的粗定位。然后对人脸候选区域采用AdaBoost算法进行精检测。该系统使用简单,检测效果较好。

A practical face detection system which is based on VS2010 and Intel open source computer vision library （OpenCV） under the platfrom of windows.was designed. It shoud bring out the face region detection in the complex backgroud image which may consist of face regions. The realization of face detection function is mainly fusion of dual skin models and Adaboost algorithm. The image segemention of skin region was firstly got based on a smiple boundary skin model and Gaussian skin color model in the YCbCr color space. The face coarse region location was determined by using the results of skin color segmentaion. Combining Adaboost algorithm, the accurate candidate face region was acquired secondly. This system is easy to use and has better detection.