Super Converging Speed of Our Nonlinear Auto-adapted Iterative Reconstructing Technique

It discussed the converging speed of a new algebraic reconstructing algorithm, our nonlinear auto-adapted iterative reconstructing technique, named Simple Self-correlative algebraic reconstruction technique (SSART). With numerical simulation, SSART was applied to reconstructing a complex simulated field in order to test its converging speed. For contrast, three current representative algebraic reconstructing techniques (ARTs) including basic algebraic reconstruction technique (ART), simultaneous ART (SART) and modified SART (MSART) were simulated in the same way. The calculated results and reconstructive accuracy were discussed with mean-square error (MSE) and peak error (PE). Based on the MSE and PE analysis, the converging speed of each reconstructing technique was discussed. As the result, the converging speed was improved much by SSART. When we iterated by the end of the sixth iterating cycle, its MSE decreased by 93.0% from that of ART at the level of 10-4 magnitude, and its PE decreased 98.4% at the level of 10-3 magnitude. SSART was considered a very superior reconstructing technique that had quicker converging speed and higher reconstructive accuracy. It´s a superior iterating reconstructing technique to our knowledge.