一种改进的加速稳健特征图像配准算法  

Improved SURF Image Registration Algorithm

作　　者：孙聪 冯孝周[1] 孙浩 张文娟 SUN Cong;FENG Xiaozhou;SUN Hao;ZHANG Wenjuan(School of Science,Xi’an Technological University,Xi’an 710021,China;Test and Measuring Academy of Norinco,Huayin 714200,China)

机构地区：[1]西安工业大学基础学院,西安710021 [2]中国兵器工业集团公司测试研究院,华阴714200

出　　处：《西安工业大学学报》2022年第6期552-558,共7页Journal of Xi’an Technological University

基　　金：国家自然科学基金(12001425);陕西省社会科学重点课题研究项目(2022HZ1809);国家大学生创新创业训练计划项目(202110702010);西安工业大学研究生教育改革重点项目(XAGDYJ220106)。

摘　　要：针对传统的加速稳健特征配准算法存在计算复杂、配准精度不高、鲁棒性差等不足,文中提出了一种改进的加速稳健特征配准算法。通过构建尺度空间及计算Hessian矩阵的极值筛选出稳定的特征点,利用二进制稳健独立基本特征描述子对特征点进行描述,并通过几何约束算法对随机一致性抽样算法进行改进,实现了匹配点对的提纯,求出单应矩阵并应用于待配准图像。研究结果表明:文中的算法相较于加速稳健特征配准算法,在噪声环境下配准偏移下降了13.8%,仿真实验中配准偏移平均下降了14.7%,配准时间平均降低了15.63%,表现出优良的抗噪性、准确性及实时性。The speeded up robust features(SURF)alignment algorithm present has deficiencies such as complicated computation,low alignment accuracy and poor robustness.An improved SURF alignment algorithm is proposed in this paper.Firstly,the stable feature points are selected by constructing the scale space and calculating the extrema of hessian matrix.Secondly,they are described by the binary robust independent elementary Features(BRIEF)descriptor.The random consistent sampling algorithm is improved by the geometric constraint algorithm,purifying the matched point pairs.Finally,the single response matrix is derived and applied to the image to be aligned.The results show that compared with the SURF alignment algorithm,by the improved method,the alignment offset decreases by 13.8%in a noisy environment and by 14.7%on average in the simulation experiments,and the alignment time is reduced by 16.75%on average.It can be concluded that the improved method shows excellent noise immunity,accuracy and real time performance.

关 键 词：图像配准 加速稳健特征 二进制稳健独立基本特征描述子 特征检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]