Optical CMOS-Based Method for Ionized Particle Identification and Extraction

摘要: The integration of visible light imaging and ionizing radiation detection has long been hindered by their distinct physical mechanisms, posing challenges for advancements in nuclear safety monitoring and high-resolution imaging technologies. In this study, we propose a novel method for the discrimination and extraction of ionizing particles based on their spatial morphological features, leveraging a CMOS active pixel sensor (APS). By systematically analyzing the morphological characteristics of ionization response events from α, β, and γ radiation under varying gain and integration time conditions, parameters such as pixel count, aspect ratio, convexity, and compactness are extracted, with a focus on their relationship with the mean pixel value. The results demonstrate that α-particle events exhibit significant differences in mean pixel value, pixel count, rectangularity, and convexity compared to β- and γ-particle events, while β- and γ-particle features show notable overlap, distinguishable primarily by mean pixel value. The proposed method effectively discriminates and extracts mixed radiation events, offering a foundation for future machine learning-based intelligent classification. This work highlights the potential of APS sensors in high-resolution optical imaging and nuclear radiation detection, providing critical insights for their applications in nuclear safety, emergency response, and facility decommissioning.