摘要
由于弱纹理图像无显著的纹理特征,使得图像特征提取不完整。为了通过图像特征进行匹配,针对弱纹理人脸图像,无法根据图像特征进行匹配,导致匹配效果差的问题,提出改进径向基算法的弱纹理人脸图像匹配优化方法。通过采集不同姿势时同一个人的脸部几何特征,得到人脸图像各个区域的平均几何特征,并对图像中人脸的旋转信息进行度量,融合径向基函数变形技术对人脸形状变形,提取弱纹理人脸图像的原始纹理信息,选择相应的模板纹理信息进行置换,得到新的纹理信息特征,并将获取的纹理特征叠加在已经变形的人脸图像上,精确地完成对弱纹理人脸图像的匹配。仿真证明,采用改进的弱纹理人脸图像匹配,可以有效减少人脸图像匹配的误差,提高图像匹配速率。
An optimization matching method of weak texture human face image based on the modified radial basis algorithm is proposed. The facial geometric feature of the same person with various poses is collected to obtain mean geometric feature of each region. Facial rotary information is measured. The facial shape to extract original texture information integrated with radial basis function is deformed. Then the relevant template texture information is chosen and replaced to obtain new information feature. The obtained feature is superimposed upon the facial image deformed already. Finally, the matching of weak texture face image is completed exactly. The simulation resuhs show that the modified method can reduce matching error of facial image effectively and improve matching rate apparently.
出处
《计算机仿真》
CSCD
北大核心
2016年第7期331-334,共4页
Computer Simulation
关键词
弱纹理
人脸图像
匹配
Weak texture
Face image
Matching
