摘要
人脸检测在图像检测领域有许多重要的应用。随着技术的普遍使用,在压缩领域直接检测将成为检测领域一项比较有意义的工作。本文综合了人脸的肤色特征、亮度、频谱特征及结构特征,提出一种在压缩域中检测人脸的新算法。根据肤色模型、数学形态学处理检测出肤色区域,基于颜色特征人脸图像准确快速地分割出脸部的非肤色区域,然后根据压缩域中对应的交流系数(频谱特征)筛选眼睛候选区域,最后模板匹配确定人脸。本算法用于基于DCT的压缩标准,运算速度比较快,准确率比较高。
Face detection is valuable in the image detection. With the wider application of compression technology, compressed--domain image process have been paid more attention. An Algorithm of face detection in compressed domain is presented, which is based on skin color,Brightness and Spectrum. First, skin regions are detected based on a model of skin color, Each block is morphologically processed in order to smooth out the biconvex and small backdate and to fill holes. The non-skin color regions are further segmented with the color features image. It can get finite candidate regions of eyes and match these regions to find the best match. The Algorithm is effective and robustness for the compressed bit-streams, which are encoded by DCT-based encoding algorithm.
出处
《电子测量技术》
2008年第8期137-139,166,共4页
Electronic Measurement Technology
关键词
人脸检测
肤色模型
压缩域
离散余弦变换
face detection
skin model
compressed domain
discrete cosine transform
作者简介
王文宁，女，1979年出生，汉族，山东烟台人，助教，硕士，主要研究方向为图像处理。E-mail：wwn812@sina．com
师磊，男，1979年出生，中级工程师，主要研究方向为，计算机技术。
