摘要
传统的Shape Context算法只对简单的形状进行形状匹配.关注于人脸形状匹配和智能化视频监控应用,提出一种Shape Context形状匹配的改进方法,并将Shape Context的形状匹配运用到人脸的形状匹配中.本方法基于Shape Context算法,使用改进的边界提取算法,并融入了扩散滤波的预处理算法和数学归一化方法.本方法适用简单形状和人脸形状匹配,具有二维不变性,并且在智能视频监控中能得到稳定和有效的使用.
Traditional Shape Context is used in simple shape matching. Focusing on the face shape matching and intelligent video surveillance applications, an improved Shape Context method is proposed, and use Shape Context in the face shape matching. Based on Shape Context, this method is a 2D invariance shape matching algorithm, integrated with improved contour extraction algorithm, pretreatment （diffusion filter algorithm） and normalization method, which make it suitable for both simple shape and face shape matching. This method can be used in intelligent video surveillance system robustly and effectively.
出处
《东华大学学报（自然科学版）》
CAS
CSCD
北大核心
2009年第1期79-83,共5页
Journal of Donghua University(Natural Science)
基金
全IP网络智能化安防系统技术研究及应用实现(065115011)
关键词
智能视频监控
形状匹配
人脸匹配
形状上下文
扩散滤波
intelligent video surveillance
shape matching
face matching
shape context
diffusion filter
