Abstract: 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.

Key words:  local binary descriptors, local invariant features, optimization of local binary descriptors, evaluation of local binary descriptors, features matching, target recognition