Improvement Of Image Registration Method Based On Fractional Fourier Transform

In the field of image processing,image registration,as one of the hot issues to be studied,inevitably has some difficulties in the research.It belongs to the image preprocessing method,and is also the preprocessing step in the fields of image recognition and image fusion,which is closely related to the success or failure of the subsequent processing technology.At present,many researchers use Fractional Fourier transform(FRFT)as a research method,and have obtained some research results in image fusion,image registration,etc.,but with the diversity of digital images,it can no longer meet the needs of modernization.In this paper,the image with translation and rotation relationship is mainly taken as the research object.Based on FRFT,the image registration is further improved,so that the image with translation and rotation relationship can be registered in the FRFT domain.The feasibility of the method is verified through simulation experiments.In this paper,two methods are proposed,one is image registration method based on FRFT and Genetic algorithm(hereinafter referred to as GA),the other is image registration method based on FRFT and GA and Particle swarm optimization algorithm(hereinafter referred to as PSO).Firstly,two-dimensional FRFT and phase correlation method are used to perform rough registration for images with translation transformation and rotation transformation,and rough registration parameters are obtained.The next step is to find a new similarity criterion based on FRFT and the target function MPSNR.For the image registration method based on FRFT and GA,the above similarity criterion is used as the fitness function of GA,and then the simulation experiment is carried out.MPSNR and mutual information were taken as fitness functions of GA respectively.For different initial population sizes,convergence algebra and running time were analyzed and compared.Then,FRFT and ga-based registration method is used to register images with translation and rotation transformation,and the optimal transformation parameters are obtained.For the image registration method based on FRFT and GA?PSO algorithm,the same similarity criterion is used as the fitness function of GA?PSO algorithm,and then the simulation experiment is carried out.The registration method based on FRFT and GA?PSO algorithm is applied to the image registration with translation and rotation transformation,and the optimal transformation parameters are obtained.Finally,the optimal transformation parameters obtained from the two new methods are compared with the coarse registration parameters and the exact values,and the error and running time are analyzed to prove the feasibility of the improved method.