摘要
提出了一种基于特征提取和验证相结合的人脸检测方法。首先应用肤色模型快速得到人脸候选区域 ,然后采用基于自适应的阈值选择策略对图像进行细分割 ,接着使用人脸椭圆形状来筛选可能的人脸 ,最后通过色度和灰度信息提取人脸特征 ,从而有效地检测出人脸。实验结果表明
Based on facial features extraction and verification, we present a new human face detection method. First, we extract face candidate regions through skin-color model. Second, with a new adaptive threshold-based region segmentation algorithm, we subseparate the image. Then, we use shape of ellipse to detect face candidates. Finally, facial features are extracted by utilizing chroma and gray information. Experimental results are satisfying.
出处
《交通与计算机》
2004年第1期29-31,共3页
Computer and Communications
关键词
人脸检测
区域分割
特征提取
特征验证
色度
人工智能
human face detection
region segmentation
features extraction
features verification
