R-L-MS-L Filter Function for CT Image Reconstruction

Abstract

In X-ray computer tomography (CT), convolution back projection is a fundamental algorithm for CT image reconstruction. As filtering plays an important part in convolution back projection, the choice of filter has a direct impact upon the quality of reconstructed images. Aim at improving reconstructed image quality, a new mixed filter based on the idea of “first weighted average then linear mixing” is designed in this article, denoted by R-L-MS-L. Here, R-L filter is relied on to guarantee the spatial resolution of reconstructed image and S-L filter is processed via 3-point weighted averaging to improve the density resolution, thus enhancing the overall reconstruction quality. Gaussian noise of different coefficients is added to the projection data to contrast the noise performance of the new and traditional mixed filters. The simulation and experiment results show that the new filter is better in anti-noise performance and produces reconstructed images with notably improved quality.