Modelling and experiments are pragmatic methods for verifying the performance of any design. Fundamentally, modelling is the first step to ensure that a design is created according to the desired characteristics. The experimental approach allows the examination of the modelling behaviour in a real environment and the verification of the design's correctness. In this research, the accuracy of an optical tomography system is evaluated by comparing the graph pattern generated through modelling and experimental procedures. Both are conducted in two different projections: single and mixed projections called the parallel beam (PB) projection and the mixed projection of parallel and fan beams left and right (MPFBLR) respectively. The evaluations confirm that the experimental and modelled patterns are consistent. This indicates that the developed optical tomography system has the desired characteristics. In this paper, a novel technique of threshold value, called filtered back projection (FBP) using averaging grouping colour (AGC), is presented. It is found that the more complex the object's location in a pipe, the higher the normalized mean square error (NMSE) and the lower the peak to noise ratio (PSNR).
