基于GPU加速的一种新图像特征匹配算法
A New Image Feature Matching Algorithm Based on GPU Acceleration
DOI: 10.12677/CSA.2019.91018, PDF,   
作者: 李 聪*, 郭大波, 刘小文：山西大学，物理电子工程学院，山西 太原
关键词: 特征检测；特征提取；BRISK；AGAST；FREAK；图像处理；Feature Detection； Feature Extraction； BRISK； AGAST； FREAK； Image Processing
摘要: 针对传统图像匹配算法匹配时间较长、误匹配较高的问题，本文提出了一种三步特征提取和匹配算法(BRISK + FREAK + BF)：首先，利用AGAST特征检测算法检测出图像的特征点；然后用FREAK描述子对特征点邻域进行描述，为特征点分配主方向和记录梯度特征；最后使用BF算法对图像的特征点进行匹配。同时，使用GPU对该算法进行加速处理。改进的算法经实验与传统的BRISK算法与FREAK算法比较分析后得知，在图像的特征匹配数目及图像特征点匹配正确率上有了一定的提高，对图像的尺度差异及亮度差异具有良好的鲁棒性，且运行速度相较于较快的BRISK算法来说提高了35%~40%左右。
Abstract: Aiming at the problem that the traditional image matching algorithm has long matching time and high mismatch, this paper proposes a three-step feature extraction and matching algorithm (BRISK + FREAK + BF): Firstly, the feature points of the image are detected by the AGAST feature detection algorithm; then using the FREAK descriptor to describe the feature point neighborhood, assigning the main direction and recording gradient feature to the feature point; finally, using the BF algorithm to match the feature points of the image, and at the same time, using the GPU to speed up the algorithm. The improved algorithm is compared with the traditional BRISK algorithm and FREAK algorithm. It is found that the number of feature matching and the correct matching rate of image feature points are improved, and have the good robustness to scale difference and brightness difference of image, and running speed is increased by 35% - 40% compared to the faster BRISK algorithm.
文章引用：李聪, 郭大波, 刘小文. 基于GPU加速的一种新图像特征匹配算法[J]. 计算机科学与应用, 2019, 9(1): 148-156. https://doi.org/10.12677/CSA.2019.91018

参考文献

[1] 吕娜, 冯祖仁. 图像匹配与跟踪研究[J]. 西安交通大学学报, 2010, 44(10): 23.
[2] 余先川, 吕中华, 胡丹. 遥感图像配准技术综述[J]. 光学精密工程, 2013, 21(11): 2960-2972.
[3] Lowe, D.G. (2004) Distinctive Image Features from Scale-Invariant Key-points. International Journal of Computer Vision, 60, 91-110. [Google Scholar] [CrossRef] 
[4] Bay, H., Tuytelaars, T. and Gool, L.V. (2006) SURF: Speeded up Robust Features. European Conference on Computer Vision, 3951, 404-417. [Google Scholar] [CrossRef] 
[5] Calonder, M., Lepetit, V., Strecha, C. and Fua, P. (2010) BRIEF: Binary Robust Independent Elementary Features. European Conference on Computer Vision, 6314, 778-792. [Google Scholar] [CrossRef] 
[6] Leutenegger, S., Chli, M. and Siegwart, R.Y. (2011) BRISK: Binary Robust Invariant Scalable Keypoints. International Conference on Computer Vision, 58, 2548-2555.
[7] Alahi, A., Ortiz, R. and Vandergh-eynst, P. (2012) FREAK: Fast Retina Keypoint. Computer Vision and Pattern Recognition, Providence, RI, 16-21 June 2012, 510-517.
[8] 石祥滨, 孙奇, 张德园, 刘芳. 一种自适应阈值分块BRISK的图像配准方法[J]. 沈阳航空航天大学学报, 2014, 31(3): 65-72.
[9] Mair, E., Hager, G.D., Burschka, D., Suppa, M. and Hirzinger, G. (2010) Adaptive and Generic Corner Detection Based on the Accelerated Segment Test. Springer, Berlin Heidelberg.
[10] 周莉莉, 姜枫. 基于FAST和BRIEF的图像匹配算法[J]. 计算机工程与设计, 2015(5): 1269-1273.
[11] 龙思源, 张葆, 宋策, 孙保基. 基于改进的加速鲁棒特征的目标识别[J]. 中国光学, 2017, 10(6): 719-725.
[12] 张恒, 刘大勇, 刘艳丽, 聂晨曦. 融合深度信息的BRISK改进算法[J]. 计算机应用, 2015, 35(8): 2285-2290.
[13] 王灿进, 孙涛, 陈娟. 基于FREAK特征的快速景象匹配[J]. 电子测量与仪器学报, 2015(2): 204-212.
[14] 叶志坚, 王福龙. 一种改进的FREAK算法的图像特征点匹配[J]. 广东工业大学学报, 2017(6): 37-42.
[15] 郑盼盼. 数据结构Brute-Force算法的实现[J]. 电子制作, 2014(6x): 85-86.
[16] 李刚. 基于GPU加速的人脸识别实现[J]. 电脑知识与技术, 2018, 14(12): 185-187.