Research on γ-photon scattering correction method based on R-SSS algorithm

As an innovative technique for industrial condition monitoring, γ-photon imaging has gained widespread development due to its excellent penetrating capability and non-contact measurement feature. Nevertheless, scattering artifacts induced by the Compton effect severely degrade image quality. To address this problem, this study proposes a Reinforced Single Scatter Simulation (R-SSS) method. This approach refines the initial scatter estimation through an adaptive convolution kernel, introduces a compensation term with high frequency to mitigate detail loss caused by convolution smoothing, and integrates an edge enhancement term to strengthen gradient response and spatial resolution in boundary areas. Subsequently, to improve the accuracy of scatterer event identification and the quality of reconstructed images, it is combined with the Ordered Subset Expectation Maximization (OSEM) reconstruction technique. To verify the effectiveness of the R-SSS algorithm in scatter correction, experiments were conducted on static Derenzo phantoms and dynamic aeroengine exhaust jet fields. Under static conditions, the images reconstructed by the proposed R-SSS algorithm achieved image quality equivalent to that of direct reconstruction while reducing imaging time by 3.98 times. Meanwhile, the integration of the R-SSS algorithm for scatter correction improved the quality of reconstructed images by 7.81% and 7.60% in static and dynamic tests, respectively, demonstrating significant advantages. In terms of scatter detection accuracy, the R-SSS algorithm outperformed the effective U-Net algorithm by 8.19% in static tests and 10.34% in dynamic experiments.