Study on Image Fusion Methods for Spatial Distribution Localization of Radiation Sources

Accurate localization of spatial distribution of radiation sources is crucial for formulating radiation protection strategies. The high-precision fusion of radioactive images and optical images of nuclear facility structures is a key procedure to guarantee the localization accuracy. In this paper, the current fusion methods, such as the weighted average method, the PCA method and the wavelet decomposition method, are comparatively investigated. Then new fusion methods are proposed due to limitations of the current methods, such as the hybrid wavelet-PCA method and the nonlinear radiation-depth information coupling method. The above fusion methods are applied to image fusion of radiation imaging and scene images of two typical scenarios. Fusion results indicate that the hybrid wavelet-PCA method significantly enhances the contrast of radiation hotspots (with an RFF reaching 0.861) while resolving color distortion issues, demonstrating overall superior performance compared to the current methods. Moreover, the radiation-depth coupling method addresses the deficiency of insufficient correspondence between radiation information and background scene elements.