Research Of 3D Image Denoising And Enhancement Algerithms Based On Fractional Calculus

Image denoising and image enhancement are two vital parts in image preprocessing. The denoised and enhanced operations of slice images are necessary for three dimensional (3D) image reconstruction. 3D image denoising can remove noises after 3D image reconstruction and increase the volume signal to noise ratio (VSNR). Effective denoising can not only add a positive influence to the reconstruction results, but also to the following analysis and processing of the reconstruction image. Most of the source slice images are not clear enough to be reconstructed. It is necessary to enhance effectively to increase the contrast and highlight some texture detail information. Based on the deep research of fractional calculus definition and the advantage of fractional integral and fractional differential, this paper has mainly finished the following work.Firstly, the principle of two dimensional (2D) fractional integral in image denoising has been studied. The theory of fractional integral in 3D edge surface tracking algorithm is also researched. The denoising algorithm with fixed order fractional integral and 3D discrete template is realized. Then the adaptive denoising algorithm is proposed. Among all the characteristics of noisy image, the gradient element is chosen to construct adaptive 3D fractional integral order. Taking advantage of the adaptive 3D fractional integral, the fractional integral order no longer needs to be manually settled. The denoising efficiency of real time is largely increased. The proposed algorithm can solve the problem which the best fractional integral order of different image is different. Adding the adaptive 3D fractional integral denoising algorithm to edge surface tracking algorithm,we can achieve the adaptive denoising goal and get higher VSNR and lower volume mean square error (VMSE).Secondly, we study the 2D fractional differential definition and its discrete template, the method to extend the 2D fractional differential calculus to 3D edge surface tracking algorithm and manually fixed order 3D fractional differential enhancement algorithm.For the sake of the shortcomings of fixed order fractional differential calculus, manually fixed 3D fractional differential can not act well to different 3D images. By means of researching the detail information and gradient information, we take the edge and texture region into consideration to propose adaptive enhancement algorithm.The maximum value of gradient variation and the maximum gradient value are selected as the adaptive fractional differential factors. Experiments results show that the advanced edge surface tracking algorithm with the proposed adaptive 3D fractional differential algorithm can reconstruct 3D slice images with higher accuracy.