Development of Machine Vision Algorithms for Radioactive Contaminated Targets Detection in Dynamic Radiation Scenarios

Detecting and monitoring radioactive contamination is very important. It ensures public safety and environmental protection. However, exploring out-of-control radioactive sources in crowded places is hard. This is true, for example, among passengers or cars. This study proposes a new approach. It is based on data fusion and machine vision methods. The approach detects radiological contamination among similar moving objects. At first, we use the motion algorithm to define 5 moving objects. They are of the same shape and size and in a two-dimensional plane. Their motion equations were inspired by the small wheeled robot. These objects move with the same speed in the plane. Next, with another algorithm based on the KLT method, we extracted related features and tracked the same objects from the image data. The algorithm combines the beam detection system's data and machine vision. It finds one or more infected targets. It successfully detects the infected moving object. This research shows a promising approach to improve monitoring of radiation environments. It suggests integrating surveillance camera images and radiation detection systems for public and large areas.