Analyzing the anti-noise performance of NAIRT used in deflection tomography with noised projections

A new iteration reconstruction technique is suggested, which is named nonlinear auto-adjusting iterative reconstruction technique(NAIRT). Its anti-noise performance used in deflection tomography was tested with its projections added noises. A complicated air flow field, called model, was simulated, and was projected according to deflection tomographic algorithm. Thereupon, the real projections were obtained. A Series of random noises at different strength level were produced using randG function. Then, the noises were added to the real projections linearly. So, a series of noised projections were acquired. According to deflection tomographic algorithm, the noised projections were inversely projected to reconstruct the model using NAMT. The reconstructive effect at the end of each cycle iteration was recorded with mean square error(MSE) index. The iteration stopped after one hundred and three cycles. As the results: First, at the noise level of 60dB S/N, NAIRT could reconstruct the model by a decent accuracy. The MSE declined to 6.49x10(-5) at the end of 103 iteration cycles. Second, at the noise level of 30dB S/N, NAIRT could yet reconstruct the model by the level of its profile. The MSE stabilized at 2.67x10(-4) at the end of 103 cycles. Last, at the noise level of 100dB SIN, NAIRT could accurately reconstruct the model. The MSE declined to 3.49x10(-5) at the end of 103 iteration cyoles. Both the reconstructed images and MSE analyses demonstrated that NAIRT had wonderful anti-noise performance when it was used in deflection tomography.