Abstract: Under the current urgent circumstances of the aviation security,all countries are intensifying the security inspection.In the view of the specialty of dangerous items,simulating the X-ray image of luggage with dangerous items is an effective way to train inspectors.This simulating method could strengthen their sensibility to the X-ray image of dangerous items in different background.In this paper,firstly analyzing the physics process behind X-ray images fusion and introducing the traditional fusing algorithm,then constructing one three layers neural network to simulate the physics process and achieving the fusion aim,in the end,comparing the neural network algorithm and traditional algorithm.It's clear that the neural network model possesses great advantages for X-ray images fusion in the precise and effect aspect.