Effect of projections number on the image quality of industrial parallel beam gamma tomography

Abstract

Gamma tomography technique can provide cross-sectional visualization of an object that needed for investigating pipe scale in geothermal power plants. Parallel beam tomography has advantages such as a simple system so that it is easier to apply in the field, but the scanning duration is relatively long. This paper discusses the effect of projections number on the reconstructed images quality and proposes the number of most effective projections for pipe scale investigation. Geothermal pipe sample (OD = 275 mm, t = 10 mm) with scale has been scanned with a parallel beam gamma tomography system. The system consists of a gamma radiation source (137Cs, 80 mCi), scintillation detector, motorized gantry, control module, data acquisition, and computer. The images were reconstructed with six different projections: 128, 64, 32, 16, 8, and 4 projections and the scanning duration: 530.8, 258.9, 127.9, 63.5, 31.7, and 15.8 minutes, respectively. Visually, the 128 projections data produces the smoothest image, whereas 64 and 32 projections images look almost the same. The 16 and 8 projections images are still able to distinguish between the pipe wall, scale, and void even though the 8 projections image looks very blurry. Then, the 4 projections image is not able to visualize the shape of the object. The gaps between the average pipe wall and void gray-scale pixels value of 128, 64, 32, 16, 8 and 8 projections images are 175, 174, 167, 153, 106, and 45, respectively. Based on the scanning duration, visualization, and the gray contrast, then the number of most effective projections is 32.