摘 要:提出一种基于面部径向曲线弹性匹配的三维人脸识别方法。使用人脸曲#br# 面上的多条曲线表征人脸曲面,提取三维人脸上从鼻尖点发射的多条面部径向曲线,对其进#br# 行分层弹性匹配和点距对应匹配,根据人脸不同部位受表情影响程度不同,对不同曲线识别#br# 相似度赋予不同权重进行加权融合作为总相似度用于识别。测试结果表明该方法具有很好的#br# 识别性能,并且对表情、遮挡和噪声具有较好的鲁棒性。#br# 关 键 词:三维人脸识别;径向曲线;分层弹性匹配;表情鲁棒性
Abstract: A 3D face recognition algorithm based on facial curves elastic matching is#br# presented in this paper. Facial surfaces are approximated with indexed sets of level curves. Firstly,#br# the radial curves emanating from the nose tip are extracted after 3D faces are preprocessed.#br# Secondly, the curves of the probe face match the curves of gallery face using the improved#br# hierarchical matching method and the two curves are matched exploiting the distance between the#br# point and nose tip. Finally, the similarities of different curves are weighted fused according to the#br# different degree influenced by expression as sum similarity which is used for matching.#br# Experiment results verify that the proposed method can obtain better recognition rate, and is#br# robust to occlusion, spikes and expression.#br# Key words: 3D face recognition; radial curves; hierarchical elastic matching; expression#br# Robustness
摘要: 提出一种基于面部径向曲线弹性匹配的三维人脸识别方法。使用人脸曲
面上的多条曲线表征人脸曲面,提取三维人脸上从鼻尖点发射的多条面部径向曲线,对其进
行分层弹性匹配和点距对应匹配,根据人脸不同部位受表情影响程度不同,对不同曲线识别
相似度赋予不同权重进行加权融合作为总相似度用于识别。测试结果表明该方法具有很好的
识别性能,并且对表情、遮挡和噪声具有较好的鲁棒性。