Effect of Filter on Image Reconstruction using Filtered Back Projection Algorithm for Industrial Gamma-Ray Tomography Technique

Abstract

Gamma-ray tomography experiment has been carried out to detect the cross-sectional spatial patterns of test objects. The resulting image quality depends on the data collection and image reconstruction process. The images were built using filtered back projection algorithm. The filter in the algorithm affects the resulting image. It is necessary to know the proper filter when reconstructing image using the algorithm. Data were collected by scanning the object using the parallel beam method. Scanning configuration was set up to every 5 mm and 32 projections (rotational scans). The scanning system consists of mechanical parts, computerized control module, a gamma-ray source (2.96 GBq of Cs-137), a NaI(Tl) scintillation detector, data acquisition and computer. In this paper, the data were reconstructed into images using back projection and also filtered back projection algorithm to study effect of the filters. The filters discussed are Ramp filter, Shepp-Logan filter, Hann filter, Hamming filter, and Cosine filter. The reconstructed images results with filter were much better than without filter. The images with no filter did not represent the object cross-sectional spatial patterns and looked blurred. There were only solid objects represented by bright white and air represented by dark colors. The images using filter could distinguish object based on its density. Ramp filtered images looked like it was filled with freckles. Shepp-Logan filter produced smoother images than Ramp filter. Hann, Hamming, and Cosine filtered images were smoother than the others. Hann and Hamming filters produced higher resolution images regarding to recognizing density value. Hann filtered images also has the smallest standard deviation. Overall, Hann filter is recommended to be used to reconstruct images from projections.