Research On Image Matching Algorithms Based On Feature Points

With the rapid development of science and technology and the continuous progress of the times,computer vision technology has become a hot topic for scholars at home and abroad,and it has been widely used in many fields,such as autopilot,intelligent monitoring,object tracking,target recognition and so on.As a key technology in computer vision,image matching is not only the premise of many image analysis and processing tasks,but also an important step in the process of oblique photogrammetry.Although there are many image matching algorithms with good real-time performance,small amount of computation,strong adaptability and high stability.However,there are still some problems,such as sensitivity to illumination changes,and the accuracy of image matching is obviously reduced in scenes with large illumination changes.Therefore,it is of great practical significance to study an image matching algorithm with better robustness against illumination changes.This paper mainly discusses how to improve the matching precision of image matching algorithms based on feature points under different illumination changing scenes.Firstly,the related principles of image matching technology are introduced.Then,from the classification of image matching algorithms,seven classical image matching algorithms based on feature points including SIFT,SURF,BRISK,ORB,FREAK,KAZE,AKAZE,are deeply studied.And three kinds of improved LBP features with strong illumination invariance,including the center-symmetric local binary pattern,the rotation invariant local binary pattern and the rotation invariant uniform local binary pattern,are introduced to the seven classical image matching algorithms to make three kind of improved image matching algorithms with better robustness against illumination changes.Finally,the simulation matching experiments under six different illumination changing scenes are designed.Through the experimental analysis,it is proved that the matching precision of the three kinds of improved algorithms in the six scenarios can be effectively improved,and the operation speed is only slightly lower than that of the original algorithm.