Abstract: Obtaining an accurate distribution map of lunar water hydrogen has significant scientific value for research on detecting water ice and for future deep space exploration. In order to effectively explore the water ice resources on the Lunar, it is necessary to accurately detect their distribution. In practical applications, a point spread function with shifting characteristics is constructed based on the detection principle of neutron detectors and the process of satellite image degradation, due to the lack of effective and reliable image sources. The point spread function is based on the kappa function. The image is blurred and noised to create a simulated detection image. Then, the maximum entropy algorithm and Richardson-Lucy algorithm are utilized to reconstruct the simulated detection image. The evaluation criteria for comparative study include visual effect, chi-square test, and authenticity test. The experimental results show that direct reconstruction cannot achieve optimal reconstruction results under both low and high levels of noise. After applying denoising preprocessing, and ensuring the safety of the chi-square test, the overall effect of the reconstruction results is found to be better than before preprocessing. Additionally, there is a significant reduction in the number of points with large deviation in the authenticity test of the reconstructed image. This indicates that the reconstruction results are now more accurate and reliable. It will provide more accurate data support for the exploration of water ice resources and deep space exploration.