摘要
针对人耳识别中对于采集角度和光照变化等条件下识别率不高的问题,提出一种基于Gabor小波和仿射尺度不变特征变换ASIFT(Affine Scale-invariant feature transform)的人耳识别方法。首先利用Gabor小波提取图像的整体特征,ASIFT提取图像特征点的局部特征,然后将整体特征与点特征联合识别,在北京科技大学提供的人耳图像库USTBⅡ中进行实验,识别率达到93%。并与其他主流人耳识别方法进行了对比实验。实验结果表明,采用基于Gabor小波和仿射尺度不变特征变换算法进行人耳图像的识别,具有较高的识别率和稳定性。
In order to solve the problem of low recognition rate in ear recognition under the conditions of the changes in acquisition angle and illumination, a new ear recognition method based on Gabor wavelet and ASIFT （affine scale-invariant feature transform） is proposed. First, Gabor wavelet is used to extract the overall features of the image, and ASIFT for extracting the local feature of image＇ s feature points ; then the overall feature and the point features are combined together for recognition. Experiments are carried out on ear image library USTB database H of Beijing University of Science and Technology, the recognition rate reaches 93%. Contrastive experiment with other mainstream ear recognition method is also conducted, the experimental results show that it can obtain a high recognition rate and stability by using the Gabor wavelet and ASIFT algorithm to recognise the ear image.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第11期202-205,共4页
Computer Applications and Software
基金
辽宁省教育厅科学研究项目(L2010194)
