摘要
人体肢体分割涉及两类最小负曲率点,利用人们对人体肢体划分的主观意识,提出最小负曲率点分析法,搜索分割手臂和腿部的最小负曲率点.跟踪目标前景区域的边缘,得到前景区域的外轮廓.采用傅里叶算子对人体轮廓实现平滑处理,消除局部噪声对最小负曲率点的影响.采用移动向量法求取轮廓曲线的数字曲率.在划分出的各个肢体部分内部进行轴线搜索,确定肢体关节点,完成人体肢体的提取.实验结果表明:算法可有效搜索最小负曲率点,实现人体肢体分割.
Human limb contour Extracting is related with two kinds of negative curvature minima (NCM). By introducing subjective judgment of the limb extracting, an algorithm of NCM is proposed to search the negative curvature minima on the limb contour. The profile of targets could be extracted by tracing their foreground. Fourier filter is implemented to smooth the profile and reduce the local noise effects on NCM. Besides, Moving vector analysis is introduced to compute the digital curvature of the profile. Also, further search of axis in local body profile is implemented to locate the joints and complete body extracting. The result of experiments showed that our algorithm provided distinct improvement in effective NCM search and human body extracting.
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2007年第5期16-18,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
关键词
图像分析
视频监控
肢体提取
最小负曲率点
image analysis
visual surveillance
limb extracting
negative curvature minima