Contrast enhancement of tomographic images by fast fourier transform filtering

Computer aided tomographic images are intrinsically affected by statistical fluctuation which degrades the contrast, spoiling thus the density resolution, a phenomenon which precludes the differentiation between neighbour regions of close densities. As this fluctuation involves frequencies usually high, when compared with those associated to the inner structure of the object under inspection, a proper filtering using a kind of low-pass filter can mitigate the problem, improving hence the density resolution. In this work, the Fast Fourier Transform – FFT has been applied to filter high frequency components of position spectra, used as projections to reconstruct 2D neutron tomographic images. These spectra, produced by a Position Sensitive Detector - PSD and by a simulator, have been filtered at different frequency cut-off, in order to evaluate its impact on the quality of the recovered images. Virtual images of a special test-object, generated by the simulator, have been used as benchmark, to assess the capability and effectiveness of the filtering procedure. These features have been corroborated by the contrast enhancement achieved for real images of different objects acquired with a neutron tomographic system equipped with a PSD. The algorithm developed to perform both simulation and filtering has been written in Fortran language