一种基于多特征融合的人脸检测技术实现

基于模板匹配的人脸检测方法,提出了两种方法来提高人脸目标的检测的精度。一种方法是利用人脸重要特征肤色,建立肤色的HSV颜色直方图模型,通过与目标区域的特征匹配,在视频序列图像中检测和定位人脸;另一种方法是利用了人脸的轮廓信息特征,建立人脸的矩特征,来解决人脸在比例、姿态和形状变化情况下的检测效率低的问题。改进的算法分别通过了人脸尺寸、形状和相似肤色实验验证,实验表明新的人脸检测方法可以有效实现对运动人脸目标的检测。

A Face Detection Based on Multi-features Fusion Technology

Template matching-based face detection method,this paper propose two methods to improve face detection accuracy target.One way is to usethe important features of human face skin color feature,the establishment of the HSV color histogram color model,with the characteristics of the targetarea matches the image,in the video sequence to detect and locate the face; Another way is to use the human face contour characteristics,to establish themoment of face features,to address the human face in proportion and shap...