| In medicine,image information fusion has a great help for doctors to understand the patient's condition.It is a very important technical means for clinical diagnosis and treatment.The image matching technology is the basis and the key of image fusion,which has a very important role in medical image analysis.Along with the progress of medical technology,the requirements for its performance are also getting higher and higher.In order to meet the high performance requirements of the medical image matching,this paper aims to improve the performance of image matching algorithm.Firstly,we present the image matching research state at home and abroad and introduce the basic theory of image,such as key elements,performance index and matching method.Then this paper explain the SIFT algorithm in detail,which is selected as the algorithm to be improved;Secondly,because there exist errors when matching,in order to satisfy the high performance requirements of medical images,this paper puts a method,which can optimize the SIFT feature matching.It is to make the graphics have affine invariance by inertia ellipse,and normalized to get the new graphics context can optimize the matching point.Thus,this improved method removes false matching points and improves the precision of the algorithm;thirdly,because the feature vector of SIFT algorithm is 128 dimensional,which takes a very long matching time,so this paper uses the image Radon transform to reduce the dimension of feature vector to 24,and use the block distance to reduce the matching time.we adope the structural similarity function to eliminate the error matching points roughly,and then take the space geometry constraints to further remove stubborn error matching points in order to achieve the matching accuracy requirements.Finally,we compare the matching effect of SIFT and improved algorithm.The result demonstrated the improved not only improve the performance but alse invariant to outside influences. |
PDF Full Text Request