Application of neural networks to baggage inspection

作者：He Mingyi; Luo Yishu; Feng Yan; Li Duanlian

来源：Neural Network World, 1997, 7(6): 631-640.

• Aviation baggage security inspection
• Computing imaging

Aviation baggage security inspection is one of the key measures for flight safety. The conventional security inspection equipment is based on an X-ray scanning principle and statistical classification methods, and there is a lot of work to be done in classification accuracy, false-alarm rate, flexibility, cost and etc. We used neural networks in classification of multi-energy X-ray images for security inspection with good results, which is believed to be the first successful neural network based X-ray security inspection. Compared with the principle in conventional X-ray scanning systems, our system, Multi-Energy X-ray inspection system with Neural Network (MEXNN), is a nonparametric classification one, overcoming the difficulties of modelling data. The neural networks we used include multilayer feedforward neural networks (MLFNN), double parallel feedforward neural networks (DPFNN) and etc. An algorithm was developed for fast learning. Experimental results have shown the MEXNN is better in performance than current similar X-ray security inspection equipment. Moreover, the cost of an MEXNN might be largely reduced. Thus it is a great social benefit and has large potential to be an economic benefit.

• 出版日期1997