局部二值描述子的研究进展综述  

Review on related studies of local binary descriptors

作　　者：边后琴 张皓霖 黄福珍 BIAN Hou-qin;ZHANG Hao-lin;HUANG Fu-zhen(School of Electronics and Information Engineering,Shanghai University of Electric Power,Shanghai 200090,China;School of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China)

机构地区：[1]上海电力大学电子与信息工程学院,上海200090 [2]上海电力大学自动化工程学院,上海200090

出　　处：《图学学报》2020年第2期254-261,共8页Journal of Graphics

基　　金：国家自然科学基金项目(61107081);上海市地方能力建设项目(15110500900)。

摘　　要：局部二值描述子是局部不变特征中的重要研究对象,广泛应用于计算机视觉与模式识别中。近年来,以BRIEF描述子为代表的局部二值描述子相继出现,对十年来局部二值描述子的研究成果与发展方向进行综述,旨在为初步研究者与工程应用人员提供参考。首先,对典型的现代局部二值描述子进行概述;其次,对优化局部二值描述子方法进行分析;最后,对相关实验评估准则进行讨论,通过总结现阶段存在的问题,给出未来研究的展望。从整体来看,近年来局部二值描述子经历了显著的发展与进步,许多对于局部二值描述子的研究均在普适性、鲁棒性和高效性上取得了成果。针对应用场景的不同,部分优化后的描述子也具备了应对实际问题的能力。这些研究进展为局部二值描述子向高层次发展、多领域拓宽打下了坚实的基础并提供了更多的思路。局部二值描述子的成功发展标志着计算机视觉技术的进步,但其发展过程中依然存在一些共性问题与矛盾,有待进一步的深入研究与解决。Local binary descriptor is an important research object in local invariant features, which is widely used in computer vision and pattern recognition. Recently, the local binary descriptors represented by BRIEF have been proposed one by one. In this paper, the research results and development of local binary descriptors in the past decade are reviewed and discussed in order to provide implications for related preliminary researchers and application engineers. Firstly, the typical modern local binary descriptors were summarized. Secondly, the methods of improving these descriptors were analyzed. Finally, the relevant experimental evaluation criteria were discussed, and the future research prospects were expounded in view of the existing problems at the present stage. As a whole, local binary descriptors have experienced remarkable development and progress in recent years, and many studies on local binary descriptors have achieved success in increasing descriptors’ universality, robustness and efficiency. Aiming at different application scenarios, some improved descriptors also have ability to deal with practical problems. Such advancement has laid a solid foundation and provided more implications for the further development of local binary descriptors characteristic of higher-level and multi-field expansion. Although the advancement of local binary descriptors marks the progress of computer vision technology, there are still some common problems and contradictions, which needs to be further studied d and solved by related researchers.

关 键 词：局部二值描述子 局部不变特征 局部二值描述子的优化 局部二值描述子的评估 特征匹配 目标识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]