【作者】 林增刚； 张艳宁； 郭哲；

【Author】 Lin Zenggang Zhang Yanning Guo Zhe (School of Computer Science and Engineering,North Western Polytechnical University,Xi’an ShanXi 710012)

【机构】 西北工业大学计算机学院；

【摘要】 相比较于二维人脸识别,三维人脸识别克服了光照、姿态等因素的影响。较大的提高了识别的准确率,但表情变化仍然影响三维人脸识别的准确率。针对这种情况,本文提出了一种新的人脸识别方法:在人脸曲面上分割出可以描述人脸特征且基本不受表情变化影响的特征曲线、区域;对曲线、区域的数据进行重新采样,保证数据量度的统一性;对重新采样后的曲线、区域进行匹配识别,以克服表情变化的影响。实验结果验证了本文方法的有效性和可行性。更多还原

【Abstract】 The three dimension face recognition has more advantages via traditional two dimension face recognition.It improves the accuracy rate for eliminating the affection of the variations of the lighting,gesture and worsen the result,but the variations of the facial expressions also affect the finally result.In order to resolve these problems,this paper bring up a new method to recognize the 3D face:first,it segment the key region & curve which is free from the facial and fully characterise the face;second, sampling again in curve and region to certify the data unit identically;fi- nally,matching the curve and region to identify 3D face.The Experiment result demonstrate the method’s availability and efficiency.更多还原