基于FAST和SURF的图像配准算法
|
摘要
尺度不变特征变换(SIFT)和加速鲁棒特征(SURF)方法在进行角点检测和特征点匹配时的时间较长。为此,提出一种改进的图像配准算法。建立参考图像与待配准图像的高斯图像金字塔,在金字塔各层图像进行检测,得到具有不同尺度的加速分割测试特征(FAST)点,采用SURF算法为各特征点分配方向,并计算各特征点的描述向量,使用快速近似最近邻搜索算法获取图像间的初始匹配点对,用随机抽样一致性算法剔除误匹配点对,同时得到2幅图像之间的几何变换矩阵。实验结果表明,与SURF算法和SIFT算法相比,该算法的特征检测速度和匹配速度较快,匹配正确率较高。
For Scale Invariant Feature Transform(SIFT)and Speeded-up Robust Feature(SURF)needing a long time in the corner detecting and feature points matching,an improved image registration algorithm is put forward.A Gaussian scale pyramid of the reference image and the matching image are established.Feature points which have different scale information are detected from each level in the image pyramid.It gets Features from Accelerated Segment Test(FAST)point with different scales.An orientation is assigned to every feature point,and feature vector is calculated by using the same way as SURF.The original matching points which have minimum Euclidean distance under some condition are determined through fast approximate nearest neighbor search.The false matching points are excluded by Randomized Sample Consensus(RANSAC) algorithm,and the transformation matrix is gained.Experimental results show that the algorithm is better than SURF and SIFT in feature detection speed and matching speed,and matching accuracy is higher.
For Scale Invariant Feature Transform(SIFT)and Speeded-up Robust Feature(SURF)needing a long time in the corner detecting and feature points matching,an improved image registration algorithm is put forward.A Gaussian scale pyramid of the reference image and the matching image are established.Feature points which have different scale information are detected from each level in the image pyramid.It gets Features from Accelerated Segment Test(FAST)point with different scales.An orientation is assigned to every feature point,and feature vector is calculated by using the same way as SURF.The original matching points which have minimum Euclidean distance under some condition are determined through fast approximate nearest neighbor search.The false matching points are excluded by Randomized Sample Consensus(RANSAC) algorithm,and the transformation matrix is gained.Experimental results show that the algorithm is better than SURF and SIFT in feature detection speed and matching speed,and matching accuracy is higher.
引文
[1]陈秀新,邢素霞.图像/视频检索与图像融合[M].北京:机械工业出版社,2012.
[2]Brown M,Lowe D G.Automatic Panoramic Image Stitching Using Invariant Features[J].International Journal of Computer Vision,2007,74(1):59-73.
[3]ZitováB,Flusser J.Image Registration Methods:A Survey[J].Image and Vision Computing,2003,21(11):977-1000.
[4]丁南南,刘艳滢,张叶,等.基于SURF-DAISY算法和随机kd树的快速图像配准[J].光电子·激光,2012,23(7):1395-1402.
[5]Lowe D G.Object Recognition from Local Scale-invariant Features[C]//Proceedings of the 7th IEEE International Conference on Computer Vision.Washington D.C.,USA:IEEE Press,1999:1150-1157.
[6]Lowe D G.Distinctive Image Features from Scaleinvariant Keypoints[J].International Journal of Computer Vision,2004,60(2):91-110.
[7]Bay H,Ess A,Tuytelaars T,et al.SURF:Speeded-up Robust Features[J].Computer Vision and Image Understanding,2008,110(3):346-359.
[8]Rosten E,Porter R,Drummond T.Faster and Better:A Machine Learning Approach to Corner Detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2010,32(1):105-119.
[9]赵璐璐,耿国华,李康,等.基于SURF和快速近视最近邻的搜索的图像匹配算法[J].计算机应用研究,2013,30(3):921-923.
[10]时颢,赖惠成,龚金辉,等.基于SURF和BBF的棉花图像匹配算法[J].江苏农业科学,2014,42(3):343-346.
[11]Lindeberg T.Scale-space Theory:A Basic Tool for Analysing Structures at Different Scales[J].Journal of Applied Statistics,1994,21(2):224-270.
[12]Muja M,Lowe D G.Fast Approximate Nearest Neighbors with Automatic Algorithm Configuration[C]//Proceedings of the 4th International Conference on Computer Vision Theory and Application.Lisboa,Portugal:INSTICC Press,2009:331-340.
[2]Brown M,Lowe D G.Automatic Panoramic Image Stitching Using Invariant Features[J].International Journal of Computer Vision,2007,74(1):59-73.
[3]ZitováB,Flusser J.Image Registration Methods:A Survey[J].Image and Vision Computing,2003,21(11):977-1000.
[4]丁南南,刘艳滢,张叶,等.基于SURF-DAISY算法和随机kd树的快速图像配准[J].光电子·激光,2012,23(7):1395-1402.
[5]Lowe D G.Object Recognition from Local Scale-invariant Features[C]//Proceedings of the 7th IEEE International Conference on Computer Vision.Washington D.C.,USA:IEEE Press,1999:1150-1157.
[6]Lowe D G.Distinctive Image Features from Scaleinvariant Keypoints[J].International Journal of Computer Vision,2004,60(2):91-110.
[7]Bay H,Ess A,Tuytelaars T,et al.SURF:Speeded-up Robust Features[J].Computer Vision and Image Understanding,2008,110(3):346-359.
[8]Rosten E,Porter R,Drummond T.Faster and Better:A Machine Learning Approach to Corner Detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2010,32(1):105-119.
[9]赵璐璐,耿国华,李康,等.基于SURF和快速近视最近邻的搜索的图像匹配算法[J].计算机应用研究,2013,30(3):921-923.
[10]时颢,赖惠成,龚金辉,等.基于SURF和BBF的棉花图像匹配算法[J].江苏农业科学,2014,42(3):343-346.
[11]Lindeberg T.Scale-space Theory:A Basic Tool for Analysing Structures at Different Scales[J].Journal of Applied Statistics,1994,21(2):224-270.
[12]Muja M,Lowe D G.Fast Approximate Nearest Neighbors with Automatic Algorithm Configuration[C]//Proceedings of the 4th International Conference on Computer Vision Theory and Application.Lisboa,Portugal:INSTICC Press,2009:331-340.