摘要
尺度不变特征变换(scale invariant feature transform,SIFT)算法是目前图像研究领域的热点,它具有良好的尺度、旋转、光照、噪声等不变特性.在特征提取方法上,利用SIFT算法提取虹膜纹理的特征向量,由于提取出来的虹膜特征向量是128维,占用内存空间大,因此提出用Harris角点对初始特征点进行筛选,选择高对比度的点作为最终的虹膜特征向量;在匹配方法上,使用街区距离进行虹膜图像特征匹配,进一步提高虹膜图像匹配的速度.实验结果表明,改进的算法在保持鲁棒性的同时,提高了SIFT特征匹配效率,能够为一些快速应用提供保障.
Scale invariant feature transform algorithm was a hot topic in the field of image research,because of its good scale,rotation,illumination,noise and other invariant features. In the feature extracting method,SIFT algorithm was used to extract the feature vector of iris texture,however the iris feature vector was 128 dimension,which occupied a large memory space. Therefore,it was proposed that the initial feature points were selected by the Harris corner points,and the point of high contrast were selected as the final iris feature vectors. In the matching method,the block distance was used to match the feature of iris image,which further improved the matching speed of iris image. The experimental results showed that the improved algorithm increased the SIFT feature matching efficiency while maintaining the robustness,and which could provided protection for some fast applications.
出处
《郑州大学学报(理学版)》
CAS
北大核心
2017年第3期14-19,共6页
Journal of Zhengzhou University:Natural Science Edition
基金
河南省科技攻关项目(152102210032)