留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于四叉树的改进BRISK特征提取算法

骆开庆 杨坤 张健 肖化

骆开庆, 杨坤, 张健, 肖化. 基于四叉树的改进BRISK特征提取算法[J]. 华南师范大学学报(自然科学版), 2020, 52(2): 114-121. doi: 10.6054/j.jscnun.2020034
引用本文: 骆开庆, 杨坤, 张健, 肖化. 基于四叉树的改进BRISK特征提取算法[J]. 华南师范大学学报(自然科学版), 2020, 52(2): 114-121. doi: 10.6054/j.jscnun.2020034
LUO Kaiqing, YANG Kun, ZHANG Jian, XIAO Hua. The Improved BRISK Feature Extraction Algorithm Based on Quadtree[J]. Journal of South China normal University (Natural Science Edition), 2020, 52(2): 114-121. doi: 10.6054/j.jscnun.2020034
Citation: LUO Kaiqing, YANG Kun, ZHANG Jian, XIAO Hua. The Improved BRISK Feature Extraction Algorithm Based on Quadtree[J]. Journal of South China normal University (Natural Science Edition), 2020, 52(2): 114-121. doi: 10.6054/j.jscnun.2020034

基于四叉树的改进BRISK特征提取算法

doi: 10.6054/j.jscnun.2020034
基金项目: 

广东省科技计划项目 2015A030401086

详细信息
    通讯作者:

    张健,讲师,Email:jianzhang@m.scnu.edu.cn

  • 中图分类号: TP391.4

The Improved BRISK Feature Extraction Algorithm Based on Quadtree

  • 摘要: 针对BRISK算法计算速度稍慢、提取的特征点容易出现扎堆的问题,利用四叉树均匀化特征点的方法,提出了基于四叉树的改进BRISK特征提取算法(Quad-BRISK算法):在生成的图像金字塔上提取并检测出具有尺度不变性的特征点之后,采用四叉树方法划分特征点,再计算特征点的方向和BRISK描述子,经过粗匹配、筛选、提纯后最终得到精匹配图像.利用Mikolajczyk和Schmid的特征对比实验图集,对SIFT、ORB、BRISK与Quad-BRISK算法进行了测试对比实验.实验结果表明:Quad-BRISK算法不仅能够提取更加稳定的特征点,同时提高了特征点的匹配精度和计算速度.
  • 图  1  FAST特征点检测[9]

    Figure  1.  The feature point detection with FAST[9]

    图  2  四叉树方法划分特征点

    Figure  2.  The partition feature points with the Quadtree method

    图  3  N=60的BRISK采样模式[4]

    Figure  3.  The BRISK sampling pattern with N=60[4]

    图  4  Oxford标准图集

    Figure  4.  The Oxford image set

    图  5  4种算法在Graf图像中的特征点匹配效果

    Figure  5.  The feature point matching effect images of four algorithms in Graf images

    图  6  4种算法在Boat图像中的特征点匹配效果

    Figure  6.  The feature point matching effect images of four algorithms in Boat images

    图  7  4种算法在Leuven图像中的特征点匹配效果

    Figure  7.  The feature point matching effect images of four algorithms in Leuven images

    图  8  4种算法在Trees图像中的特征点匹配效果

    Figure  8.  The feature point matching effect images of four algorithms in Trees images

    图  9  4种算法在UBC图像中的特征点匹配效果

    Figure  9.  The feature point matching effect images of four algorithms in UBC images

    表  1  4种算法在图像仿射变化下特征点的匹配正确率

    Table  1.   The correct matching rate of feature points in four algorithms under image affine changes %

    组号 SIFT ORB BRISK Quad-BRISK
    1 82.39 90.75 82.03 92.78
    2 67.18 72.82 66.45 91.43
    3 - - - -
    4 - - - -
    5 - - - -
    注:“-”表示特征点匹配正确率过低或者算法失效,无意义.
    下载: 导出CSV

    表  2  4种算法在图像尺度和旋转变化下特征点的匹配正确率

    Table  2.   The correct matching rate of feature points in four algorithms under image scale and rotation changes %

    组号 SIFT ORB BRISK Quad-BRISK
    1 79.44 93.07 85.11 91.57
    2 77.37 91.54 69.86 100
    3 60.56 82.18 - -
    4 - 77.03 - -
    5 - - - -
    注:“-”表示特征点匹配正确率过低或者算法失效,无意义.
    下载: 导出CSV

    表  3  4种算法在图像光照变化下特征点的匹配正确率

    Table  3.   The correct matching rate of feature points in four algorithms under image lighting changes %

    组号 SIFT ORB BRISK Quad-BRISK
    1 80.56 88.77 85.35 95.17
    2 79.97 81.39 78.13 93.39
    3 76.25 78.66 71.02 90.32
    4 70.47 79.13 65.24 94.59
    5 65.49 77.39 - 93.72
    注:“-”表示特征点匹配正确率过低或者算法失效,无意义.
    下载: 导出CSV

    表  4  4种算法在图像模糊变化下特征点的匹配正确率

    Table  4.   The correct matching rate of feature points in four algorithms under image blur changes %

    组号 SIFT ORB BRISK Quad-BRISK
    1 80.25 91.43 87.76 88.81
    2 - 90.75 84.41 92.24
    3 - 81.37 60.68 91.14
    4 - 69.90 - 80.30
    5 - - - 84.09
    注:“-”表示特征点匹配正确率过低或者算法失效,无意义.
    下载: 导出CSV

    表  5  4种算法在图像压缩下特征点的匹配正确率

    Table  5.   The correct matching rate of feature points in four algorithms under image compression %

    组号 SIFT ORB BRISK Quad-BRISK
    1 85.68 99.67 97.50 99.46
    2 80.26 99.44 95.34 98.97
    3 75.62 98.31 91.01 98.12
    4 65.32 95.57 80.52 97.43
    5 57.78 93.17 59.27 93.37
    下载: 导出CSV

    表  6  4种算法的运行时间分析

    Table  6.   The analysis of the running time of four algorithms ms

    图集 SIFT ORB BRISK Quad-BRISK
    Graf 469.9 29.16 121.9 83.89
    Boat 564.4 39.38 144.0 113.80
    Trees 746.6 56.16 224.7 154.90
    Leuven 495.2 27.12 118.0 76.47
    UBC 498.2 32.82 120.6 98.95
    下载: 导出CSV
  • [1] LOWE D G. Distinctive image features from scale-invariant keypoints[J]. International Journal of Computer Vision, 2004, 60(2):91-110. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=6d7f4db9dae5b32514309ee242a50f4b
    [2] BAY H, ESS A, TUYTELAARS T. SURF:speeded up robust features[J]. Computer Vision & Image Understan-ding, 2008, 110(3):346-359 http://d.old.wanfangdata.com.cn/Periodical/hwyjggc200901035
    [3] RUBLEE E, RABAUD V, KONOLIGE K, et al. ORB: An efficient alternative to SIFT or SURF[C]//Proceedings of the 2011 International Conference on Computer Vision. Barcelona, Spain: IEEE, 2012: 2564-2571.
    [4] LEUTENEGGER S, CHLI M, SIEGWART R Y. BRISK: binary robust invariant scalable keypoints[C]//Procee-dings of the 2011 International Conference on Computer Vision. Barcelona, Spain: IEEE, 2011: 6-13.
    [5] 黄钰雯, 胡立坤, 卢泉, 等.基于改进SURF-BRISK算法的航拍图像拼接方法[J].广西大学学报(自然科学版), 2017, 42(3):1058-1068. http://d.old.wanfangdata.com.cn/Periodical/gxdxxb201703030

    HUANG Y W, HU L K, LU Q, et al. A mosaic method of unmanned aerial vehicle images based on the improved SURF-BRISK[J]. Journal of Guangxi University(Natural Science Edition), 2017, 42(3):1058-1068. http://d.old.wanfangdata.com.cn/Periodical/gxdxxb201703030
    [6] 赵婷, 康海林, 张正平.结合区域分块的快速BRISK图像拼接算法[J].激光与光电子学进展, 2018, 55(3):210-215. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jgygdzxjz201803024

    ZHAO T, KANG H L, ZHANG Z P. Fast image mosaic algorithm based on area blocking and BRISK[J]. Laser & Optoelectronics Progress, 2018, 55(3):210-215. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jgygdzxjz201803024
    [7] CEN C, LI R Q, XU X. Fast robust matching algorithm based on BRISK and GMS[J]. Journal of Physics:Conference Series, 2019, 1237(2):022070/1-6.
    [8] MUR-ARTAL R, MONTIEL J M M, TARDOS J D. ORB-SLAM:a versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(5):1147-1163. http://d.old.wanfangdata.com.cn/Periodical/jsjyy201705041
    [9] ROSTEN E, DRUMMOND T. Machine learning for high-speed corner detection[C]//Proceedings of the 9th European Conference on Computer Vision. Berlin: Springer, 2006: 430-443..
    [10] KLEIN G, MURRAY D. Improving the agility of keyframe-based SLAM[C]//Proceedings of the 10th European Conference on Computer Vision. Berlin: Springer, 2008: 802-815.
    [11] ROSIN P L. Measuring corner properties[J]. Computer Vision and Image Understanding, 1999, 73(2):291-307. http://d.old.wanfangdata.com.cn/NSTLQK/NSTL_QKJJ023455194/
    [12] 成怡, 佟晓宇.基于改进ORB算法的移动机器人视觉SLAM方法研究[J].电子技术应用, 2019, 45(1):10-13;18. http://d.old.wanfangdata.com.cn/Periodical/dzjsyy201901003

    CHENG Y, TONG X Y. Research on visual SLAM method of mobile robot based on improved ORB algorithm[J]. Application of Electronic Technique, 2019, 45(1):10-13;18. http://d.old.wanfangdata.com.cn/Periodical/dzjsyy201901003
    [13] HARRIS C G, STEPHENS M. A combined corner and edge detector[C]//Proceedings of the Fourth Alvey Vision Conference. Manchester: [s.n.], 1988: 147-151.
    [14] TRAN Q H, CHIN T J, CARNEIRO G, et al. In defence of RANSAC for outlier rejection in deformable registration[C]//Proceedings of the 12th European Conference on Computer Vision. Berlin: Springer, 2012: 274-287.
    [15] MIKOLAJCZYK K, TUYTELAARS T, SCHMIS C, et al. A comparison of affine region detectors[J]. International Journal of Computer Vision, 2005, 65:43-72. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=5b49a81863c6d17cdfba4efb1ddef11b
    [16] 朱成德, 李志伟, 王凯, 等.基于改进网格运动统计特征的图像匹配算法[J].计算机应用, 2019, 39(8):2396-2401. http://d.old.wanfangdata.com.cn/Periodical/jsjyy201908034

    ZHU C D, LI Z W, WANG K, et al. Image matching algorithm based on improved RANSAC -GMS[J]. Journal of Computer Applications, 2019, 39(8):2396-2401. http://d.old.wanfangdata.com.cn/Periodical/jsjyy201908034
  • 加载中
图(9) / 表(6)
计量
  • 文章访问数:  2606
  • HTML全文浏览量:  1313
  • PDF下载量:  56
  • 被引次数: 0
出版历程
  • 收稿日期:  2019-10-29
  • 刊出日期:  2020-04-25

目录

    /

    返回文章
    返回