Object recognition based on foreground detection using X-ray imaging

Development of effective terrorism-prevention techniques has assumed international importance since the 9/11 attacks. X-ray examination technology is a nondestructive screening method that enables the contents and internal structures of objects to be viewed without destroying the packaging or the external structures. Owing to its speed and efficiency compared to manual inspection, this technology is especially useful for the inspection of checked luggage and cargo, especially for inspecting items that cannot be disassembled. This paper presents developed training software for automated X-ray image recognition for checked luggage. To obtain the optimum classification result, an algorithm is designed using the C++ language. In Taiwan, the intensity of X-ray equipment is 10 times that of a typical X-ray baggage inspection machine. The use of reflected thermal neutrons in this equipment can reveal the presence of drugs, weapons, or explosives that are hidden in luggage. Using the Harris corner detection method and the speeded-up robust features (SURF) algorithm, the corner and feature points are obtained. In an evaluation, the proposed system uses a pattern-matching method and SURF to detect a 9-mm pistol. Compared to the results of a gun template image, the similarity obtained by the proposed method is increased by 0.85 to 1.81%.