Abstract: In order to solve the problem of image denoising in irradiation fields such as nuclear waste disposal hot cells, this study made corresponding improvements to the algorithm based on the idea of median filtering for radiation noise points, and finally proposed an algorithm suitable for image denoising in strong irradiation fields. The algorithm process is divided into three steps: noise recognition, noise processing and detail protection. Noise recognition starts from visual perception. If the pixel value is higher than its surrounding value and reaches the set threshold, it is marked as suspicious. By setting the maximum area threshold of radiation noise, the area larger than the threshold is excluded. The idea of median filtering is used to deal with noise, and the size of filtering window is adjusted adaptively. The protection of detail borrows from the idea of image fusion. Compared with other noise reduction algorithms, the results show that the proposed algorithm is accurate in the identification of radiated noise points, has obvious noise reduction effect, and protects the details of the picture to the greatest extent. The algorithm demonstrates good robustness under radiation environments with different dose rates, and still achieves effective denoising even under a strong irradiation field of 100 Gy/h. This can provide technical support for intelligent applications in nuclear waste disposal.