[1]张娓娓,赵金龙,何 佳,等.一种基于阶阵列的 BRIEF 特征描述子[J].计算机技术与发展,2023,33(05):81-87.[doi:10. 3969 / j. issn. 1673-629X. 2023. 05. 013]
ZHANG Wei-wei,ZHAO Jin-long,HE Jia,et al.BRIEF Feature Descriptor Based on Order Array[J].,2023,33(05):81-87.[doi:10. 3969 / j. issn. 1673-629X. 2023. 05. 013]
点击复制
一种基于阶阵列的 BRIEF 特征描述子(
)
《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]
- 卷:
-
33
- 期数:
-
2023年05期
- 页码:
-
81-87
- 栏目:
-
媒体计算
- 出版日期:
-
2023-05-10
文章信息/Info
- Title:
-
BRIEF Feature Descriptor Based on Order Array
- 文章编号:
-
1673-629X(2023)05-0081-07
- 作者:
-
张娓娓1 ; 赵金龙1 ; 何 佳1 ; 陈绥阳1; 2 ; 王 杰1
-
1. 西安思源学院 电子信息工程学院,陕西 西安 710038;
2. 西安交通大学 理学院,陕西 西安 710071
- Author(s):
-
ZHANG Wei-wei1 ; ZHAO Jin-long1 ; HE Jia1 ; CHEN Sui-yang1 ; 2 ; WANG Jie1
-
1. School of Electronics and Information Engineering,Xi’an Siyuan University,Xi’an 710038,China;
2. School of Science,Xi’an Jiaotong University,Xi’an 710071,China
-
- 关键词:
-
描述子; 二值模式; BRIEF; 阶排列; 旋转不变性
- Keywords:
-
descriptors; binary pattern; BRIEF; order array; rotation invariance
- 分类号:
-
TP391
- DOI:
-
10. 3969 / j. issn. 1673-629X. 2023. 05. 013
- 摘要:
-
局部特征匹配是机器视觉研究领域中的一个基础问题,也是该领域的研究热点之一,在目标识别、目标跟踪、场景区分等应用中具有重要的作用。 而在局部特征匹配研究过程中,如何在满足多种图像变换的前提下,设计一种高效的图像特征描述子是需要解决的一个关键问题。 现有的特征描述子,如 SIFT 和 SURF,计算复杂性较高,难以胜任实时视频或移动计算环境;BRIEF 特征描述子计算简单,匹配效率高,能满足实时视频或者移动计算环境的要求,但其仅考虑了单个像素,不具备方向,也就不具有旋转不变性。 在 BRIEF 特征描述子的基础上,该文选择多个特征点,并引入阶排列方法,提出一种改进的特征描述子 OPoBRIEF。 相对于传统的特征描述子,OPoBRIEF 能够包含更多的局部特征信息,并且计算复杂性较低。 通过特征描述子稳定性实验,表明 OPoBRIEF 比 BRRIEF 具有更高的匹配正确率和更好的稳定性。 而特征描述子旋转不变性的实验则表明,在旋转角度为 10 ~ 12 区间,OPoBRIEF 与 SIFT 效果相当,但明显优于 ORB 算法。
- Abstract:
-
Local feature matching is a basic and hot problem in the area of computer vision, which plays an important role in manyapplication fields such as object recognition,visual tracking,scene classification and so on. The key to this problem is how to design aneffective image feature descriptor regarding for different image deformations in the process of local feature matching research. Existingfeature descriptors,such as SIFT and SURF,lack the ability for real-time or mobile applications due to their high complexity. BRIEFfeature descriptor is simple in calculation and has high matching efficiency,which can meet the requirements of real-time video or mobilecomputing environment. However,it only considers a single pixel and does not have direction,so it does not have rotation invariance.Based on the BRIEF, we select several feature points and introduce the order arrangement method to propose an improved featuredescriptor OPoBRIEF. Compared with traditional feature descriptors,OPoBRIEF can contain more local feature information and has lowercomputational complexity. The feature descriptor stability experiments show that OPoBRIEF has higher matching accuracy and betterstability than BRRIEF. The experiment on the rotation invariance of feature descriptors shows that OPoBRIEF has the same effect as SIFTin the rotation angle range of 10 ~ 12,but it is obviously better than ORB algorithm.
更新日期/Last Update:
2023-05-10