摘要
提出一种基于Gabor相位与局部二值模式(local binary patterns,LBP)算子的活动表观模型(activeappearance model,AAM)。与基于亮度的AAM相比,改进模型在三个方面提高了算法性能:提供多尺度多方向的Gabor纹理,提高了模型的匹配精度;增强了对外部环境变化(如光照)的鲁棒性;基于LBP的纹理编码去除了大量冗余。实验结果表明该模型能够有效提高模型的匹配精度。
An active appearance model（AAM） based on Gabor phase and local binary patterns（LBP） is presented.In comparison with the intensity based AAM,the proposed model improves the performance in three aspects： providing multi-scale and multi-direction Gabor texture which enhances the fitting accuracy of the model;improving the robustness to environmental changes,e.g.,illumination;LBP based texture coding reduces a large amount of the redundency.Experimental results on various datasets demonstrate that the proposed model can effectively improve the fitting accuracy.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2010年第5期1051-1054,1104,共5页
Systems Engineering and Electronics
基金
国家自然科学基金(60702061)资助课题
