摘要
This paper reviews and evaluates several state-of-the-art feature description algorithms. The components of each feature description method are analyzed and their applications in dealing with specific challenges are identified. In the paper, we compare state-of-the-art feature description methods including the SIFT, DAISY, HRI-CSLTP, LIOP, MROGH and MRRID with specific measurement regulation. The quantitative comparative results demonstrate these algorithms' applications in different scenes, which provide a certain guide for designing novel feature description algorithms.
- 出版日期2015
- 单位中国科学院光电技术研究所; 中国科学院大学