Abstract:
Aiming at the problem that traditional Scale Invariant Feature Transform(SIFT) algorithm may cause error match in low resolution palmprint image, this paper proposes a novel approach of palmprint recognition based on improved SIFT algorithm. According to the thought of local match, the algorithm combines the Euclidean distance and weighted sub-region. The similarity calculated can reflect the local and global features of images. Simulation results prove that the recognition rate of the improved SIFT algorithm is higher than the recognition rate of SIFT algorithm.
Key words:
Scale Invariant Feature Transform(SIFT),
Euclidean distance,
weighted sub-region palmprint recognition,
illumination invariance
摘要: 针对在低分辨率掌纹图像中,传统尺度不变特征转换算法易产生误匹配的问题,提出一种用于掌纹识别的改进尺度不变特征转换算法。根据局部匹配的思想,结合欧氏距离及加权子区域匹配方法对图像进行匹配,计算得出的相似度能反映图像的局部与全局特征。仿真实验结果证明,改进的尺度不变特征转换算法比原算法具有更高的识别率。
关键词:
尺度不变特征转换,
欧氏距离,
加权子区域,
掌纹识别,
光照不变性
CLC Number:
JI Zhong, WANG Zheng-Yong, FU Li, ZHANG Shen-Wei, HE Jun. Improved Algorithm of Scale Invariant Feature Transform in Palmprint Recognition[J]. Computer Engineering, 2011, 37(17): 169-171.
瞿中, 王正勇, 傅力, 张震玮, 何军. 掌纹识别中的尺度不变特征转换改进算法[J]. 计算机工程, 2011, 37(17): 169-171.