Sharpening enhancement of Computed Tomography (CT) images using Hyperbolic Secant Square filter

In Computed Tomography (CT) imaging, selection of reconstruction kernels determines the tradeoff between image sharpness and pixel noise. This paper presents a new High Pass (HP) filter to sharp the acquired CT images without unnatural pronounced edges. Sensitivity to noise and limited directions are the major drawbacks of conventional HP filters used in image sharpening. To combat with these difficulties a two Dimensional (2D) isotropic Hyperbolic Secant Square (HBSS) filter is developed. The 2D non-separable property of this filter improves the directional selectivity while the HBSS provide less noise sensitivity and least square design present the regularization. The improved HP filter responses are used to sharp the CT images. The performance of proposed sharpening method is compared with common unsharp mask, which uses non-separable isotropic Laplacian of Gaussian (LoG).