计算机科学 ›› 2015, Vol. 42 ›› Issue (4): 311-315.doi: 10.11896/j.issn.1002-137X.2015.04.064

• 图形图像与模式识别 • 上一篇    下一篇

一种基于分块匹配的SIFT算法

邹承明,徐泽前,薛 栋   

  1. 武汉理工大学计算机科学与技术学院 武汉430070,武汉理工大学计算机科学与技术学院 武汉430070,武汉理工大学计算机科学与技术学院 武汉430070
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受中央高校基本科研业务费专项资金(135210008),中央高校基本科研业务费专项基金(2014-VII-027),国家自然科学基金(51179146),湖北省自然科学基金(2011CDB254)资助

SIFT Algorithm Based on Block Matching

ZOU Cheng-ming, XU Ze-qian and XUE Dong   

  • Online:2018-11-14 Published:2018-11-14

摘要: SIFT算法在图像处理领域具有独特的优势,但是经过不断发展,SIFT算法在特征匹配过程中仍然具有数据处理量大、计算速度慢的问题。基于这些问题,提出了一种基于分块匹配的新型 SIFT匹配算法,它通过剔除非重叠区域来降低特征提取和匹配的时间损耗。对于图像的刚性变换,算法的核心在于图像块的切分和重叠区域的计算,首先选取少量的种子点来估算两幅图像的相关变换矩阵;然后将原始图像切分为几块,通过变换矩阵找出在匹配图中的相关块;再检测所有的匹配块上的特征点;最后结合 RANSAC算法去除伪匹配点对,来提高匹配的准确率。实验结果表明:与标准SIFT算法相比,基于分块匹配的SIFT算法在实时性和鲁棒性方面得到了进一步的提升,在实际图像匹配中具有一定的应用价值。

关键词: 分块匹配,SIFT,鲁棒性,RANSAC,变换矩阵

Abstract: SIFT algorithm has distinctive advantages in the field of image processing.However,with the development of the SIFT algorithm,it still has some disadvantages such as the large amount of data processing,slow computing speed.To address these issues,a SIFT algorithm based on block matching was proposed.It reduces the time of feature extraction and matching by extracting the overlapping areas.For the rigid transformation of a image,the core of the algorithm is to calculate the image block segmentation and overlapping areas.In the first step, a small number of seed points are selected to estimate the associated transformation matrix of two images.Then,the original image is cut into pieces and the relevant block is found by the transformation matrix.In the second step,all of the matching feature points are detected on the block.Finally,RANSAC algorithm is used to remove error matching points to improve the matching accuracy.The experimental results show that the improved SIFT algorithm of block matching has better real-time and robustness than the standard SIFT algorithm,and it has a certain application value in the actual image matching.

Key words: Block matching,SIFT,Robustness,RANSAC,Transformation matrix

[1] Lowe D G.Object Recognition from Local Scale invariant Features[C]∥Proceeding of the Seventh IEEE International Conference on Computer Vision.Kerkyra,Greece,1999:1150-1157
[2] Lowe D G.Distinctive Image Features from Scale invariant Keypoints[J].International Journal of Computer Vision,2004,60(2):91-110
[3] Ke Y,Sukthankar R.PCA—SIFT:a More Distinctive Representation for Local Image Descriptors[C]∥Proceedings of the Conference on Computer Vision and Pattern Recognition.Washington,USA,2004:511-517
[4] 吴建,马跃.一种改进的SIFT算法[J].计算机科学,2013,0(7):270-272
[5] 安建妮,刘贵喜.利用特征点配准和变换参数自动辨识的图像拼接算法 [J].红外与激光工程,2011,40(3):564-569
[6] 何婷婷,芮建武,温腊.CPU-GPU协同计算加速ASIFT算法[J].计算机科学,2014,1(5):14-19
[7] 姜超,耿则勋,娄博,等.基于GPU的SIFT特征匹配算法并行处理研究[J].计算机科学,2013,0(12):295-297,7
[8] 王晓华,傅卫平,梁元月.提高SIFT特征匹配效率的方法研究[J].机械科学与技术,2009,2(9):1252-1260
[9] Umeyama S.Least-squares estimation of transformation para-meters between two point patterns[J].IEEE Trans.Pattern Anal.Mach.Intell.,1991,13(4):376-380
[10] Fischler M A,Bolles R C.Random Sample Consensus:A Paradigm for Model Fitting with Application to Image Analysis and Automated Cartography [J].Communications of the ACM,1981,24(6):381-395

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!