Research And Application Of Adaptive Fractional Order Active Demons Algorithm

Image registration is the basic research in the field of computer vision and image processing.Most methods about the traditional image registration are mainly based on integral order calculus,which is not good for image registration with uniform grayscale,weak edge and weak texture.The advantage of fractional calculus is that it can not only enhance the weak edge and texture region of the image,but also retain the information of uniform gray area.However,in the existing methods of image registration based on fractional calculus,such as Active Demons algorithm based on fractional differential,the optimal fractional differential order needs to be manually selected through many experiments,which is easy to cause time-consuming and laborious problems.3D medical images contain abundant information,which can help doctors to make accurate judgment on their condition,but in medical image registration,The existing medical image registration methods with different modes and scales lack robustness and fall into local extremum easily.Therefore,this paper mainly focuses on the two issues,the specific work is as follows:(1)Firstly,we analyze the role of fractional calculus and select an appropriate order range of fractional calculus.Then,the gradient modulus and information entropy of the image are taken as independent variables,the inverse tangent function is selected as the prototype function,and an adaptive fractional calculus mathematical model is constructed.The mathematical model relates the order of fractional calculus to the local information of the image,so that the best fractional order of each pixel can be automatically calculated according to the local information of the image.And the model is applied to Active Demons algorithm based on fractional differential.Finally,In order to reduce the time of image registration,multi-resolution strategy is introduced to Active Demons algorithm based on fractional differential.Theoretical analysis and experimental results show that,Active Demons algorithm based on multi-resolution and adaptive fractional differential is effective in image registration with uniform grayscale,weak edge and weak texture,and the accuracy and efficiency of registration are improved to some extent.(2)3D medical images contain abundant information,Medical images of different modalities contain information about the structure of the organs and somephysiological functions,which can provide detailed information for doctors to make accurate diagnosis.In addition,different scales of medical images can also provide doctors with a basic organizational structure.Then,we studied the existing methods of3 D medical images registration and proposed the shortcomings of these methods.They lack sufficient robustness and are prone to fall into local optimum.Finally,Active Demons algorithm based on multi-resolution and adaptive fractional differential is applied to 3D medical image registration,and the validity of this algorithm is verified on two open medical image databases.Theoretical analysis and experimental results show that,Active Demons algorithm based on multi-resolution and adaptive fractional differential is effective in 3D medical image registration.