摘要
视频流中检测到的关键帧图像包含了足够的表情信息,为了将这些表情信息进行分类和识别,文章提出了一种新的弹性模板匹配算法,它首先针对经Gabor小波变换后的表情模板,运用模板图像中表情关键点的检测算法,根据表情关键点的特征信息,构造表情弹性图,通过改变表情模板弹性图中关键点的位置,将表情模板与被测表情弹性图进行非刚性匹配,进而得到两者之间的相似程度,最后通过改进的K-近邻分类策略,实现被测图像表情的有效分类与识别.
Video streaming in the key frame detected the expression of the information contained enough to classify these expressions and identifying information,a new flexible template matching algorithm is proposed,which first of all by the Gabor wavelet transform for the expression template,the template image using the key points of face detection algorithm,according to the characteristics of expression of the key points of information,expression of elastic graph constructed by changing the expression of key points in the elastic graph template position the template and measured the expression of non-rigid elastic graph expression match,and further to the degree of similarity between the two,and finally by improving the K-nearest neighbor classification strategy to achieve the effective expression of the measured image classification and recognition.
出处
《太原师范学院学报（自然科学版）》
2011年第4期87-90,共4页
Journal of Taiyuan Normal University:Natural Science Edition
基金
江苏省教育厅江苏高校自然科学研究计划项目(08KJB520001)
关键词
图像处理
弹性模板
GABOR小波变换
表情识别
image processing
flexible templates
Gabor wavelet transform
expression recognition
