Research on Improvement of Filtering Process in Filtered Backprojection Reconstruction Algorithm

Filtered back projection reconstruction algorithm is commonly used in CT image reconstruction algorithm, in which the filtering process has a great impact on the quality of image reconstruction. Aiming at the problem of low reconstruction precision and long operation time on the present commonly used filter, the existing filter is improved in this paper. The hybrid filter R-L&W-Lewit is designed which is an average of near 3 points on Lewit filter and is weighted with the R-L filter. Not only does the side-lobe oscillation of the filter decrease, but the main lobe height also increases. In this paper, we also study the feasibility for the filtered back projection algorithm by using Hartley transform to replace the Fourier transform, and use the fast Hartley algorithm to calculate the filter convolution process. The experimental results show that the new filter has high precision and good noise suppression and the effect is better than that the traditional filter. The fast Hartley algorithm can obtain similar reconstruction precision with the traditional fast Fourier algorithm, and the speed of operation has been improved obviously.