Title :
SIFT-NMI Algorithm for Image Matching
Author :
Xu, Gang ; Ma, Chen
Author_Institution :
Sch. of Electr. & Electron. Eng., North China Electr. Power Univ., Beijing, China
Abstract :
SIFT-NMI algorithm is proposed for image matching based on SIFT (Scale-invariant feature transform) and NMI (Normalized Moment of Intertia) algorithm in this paper. Firstly, the SIFT algorithm is used to obtain the coordinates and vector matrix of the image´s feature points. Then, the moment of intertia of the vector is obtained based on NMI algorithm and the pairs of matching features points are determined via setting the threshold. Ultimately, the best pairs of feature points are selected via ant colony. Matlab simulation results show that the SIFT-NMI algorithm overcomes the defects of the SIFT algorithm and NMI algorithm, since SIFT algorithm cannot do quantitative calculation and NMI algorithm cannot calculate affine image; by using SIFT-NMI algorithm we can improve the accuracy and speed of matching.
Keywords :
feature extraction; image matching; matrix algebra; vectors; Matlab simulation; SIFT-NMI algorithm; affine image; ant colony; image feature points; image matching; normalized moment of intertia; scale-invariant feature transform; threshold; vector matrix; Accuracy; Algorithm design and analysis; Clustering algorithms; Computers; Image matching; Kernel; Transforms;
Conference_Titel :
Control, Automation and Systems Engineering (CASE), 2011 International Conference on
Conference_Location :
Singapore
Print_ISBN :
978-1-4577-0859-6
DOI :
10.1109/ICCASE.2011.5997537
