The concrete-filled steel tube structure has void defects owing to factors such as construction technology, concrete shrinkage, and overload bearing, which reduce the load-bearing capacity of the structure. Furthermore, the current thermal imaging based void detection technology has a small detection depth, poor processing algorithm rendering effect, inability to detect quantitatively, and low detection efficiency. In this study, an active thermal imaging technology was used to study the void defects of concrete-filled steel tubes. A special induction heating power supply and heating probe are designed to increase the detection depth, and a heat source inversion algorithm eliminates the heating non-uniformity interference. In addition, a defect-free simulation model is established to predict the temperature distribution of the concrete-filled steel tube surface. Moreover, the difference between the experimental data and the simulation prediction data was obtained to extract the characteristics of the void defect and optimize the detection parameters to realize a quantitative detection solution. Through repeated experiments, it was shown that void defects in concrete-filled steel tubes with a wall thickness of 20mm can be detected, and the shape and size of the defects can be determined, which effectively improves the inspection depth and efficiency.