Research On The Security Detection Method Of Multi-Information Fusion Postal Sorting System

It is difficult to rely on only one kind of security detection equipment to truly realize comprehensive multi-information detection in many public situations.At present,domestic or foreign countries still cannot effectively realize the effective classification of the substances in the tested object with a complete set of technologies to comprehensively control the safety of an area.This subject proposed and designed a targeted multi-information integration of the overall scheme according to its key control of the contraband and customs used to the security mode and requirements through the investigation of national functional departments.We collected information and reconstructed the multi-information data of several common contraband,and completed a comprehensive test method through a series of theoretical and experimental studies.The equipment used to detect contraband must be able to identify the characteristics of objects in packages or containers.Some contraband such as plastic explosives can be made into many shapes or placed in many objects.This kind of object cannot be judged as contraband simply by its appearance information.This requires a device capable of identifying objects at the molecular or atomic level.X-ray-devices have been demonstrated to be capable of revealing the molecular or atomic properties of matter.X-ray techniques can provide some important properties of the substance that makes up an object.The most useful information is the density(p)of the object and the effective atomic number(Zeff).Theoretically,given the density of the object and the effective atomic number,the type of matter can be accurately determined.Studies have shown that a variety of X-ray detection techniques can be applied to the detection of contraband,but there isn't any X-ray detection technology can provide two parameters needed to accurately identify the material type of an object alone.First of all,the high energy and low energy grayscale of the substance were obtained by the double energy X-ray transmission experiment.Based on this,the mathematical model of material classification and recognition boundary curve is established.The material is initially divided into three categories:organic,inorganic and mixture.Secondly,based on the feature plane idea of extraction,a value R related to effective atomic number is extracted through dual-energy X-ray transmission experiment.This process can separate organic matter from inorganic matter and mixture,but harmless and prohibited organic matter cannot be separated because of the difference is the density.Then we studied the effect of changing external parameters such as thickness on R value,and optimized the algorithm to extract R value.The evaluation of the improved algorithm shows that the misjudgment rate of material classification is greatly reduced.Then we built the gray scale model of the scattered images with the LS algorithm combining the front scattering and back scattering images.Then the low energy scattering image and the dual-energy transmission image are combined to obtain a method to extract the characteristic value L related to object density.The influence of factors such as the Angle of package placement on L value is analyzed and reduced.The discriminant function,decision surface equation and classification discriminant rule of bayesian decision theory based on the minimum error probability are given by the material characteristic values R and L.A more effective method for material recognition is obtained by combining the dual-energy X-ray transmission technique with the low-energy scattering technique.Furthermore,the ability of X-ray detection is improved comprehensively.Third,calculating the R and L value of the object requires the real gray level of the object.The real gray level is the level of gray measured when an object is not disturbed by other background objects.Some of the objects in the package are numerous,and the objects in the package are positioned in any direction and are shielded from each other.The combination of contraband and harmless material makes detection difficult.Therefore,it is very important to remove the occlusion effect in the process of identifying material properties for getting the actual gray scale of the object.We turn the problem of n objects overlapping into the problem of two objects overlapping and focuses on the real grey scale problem of two objects overlapping.In the end,we will get the mathematical model for solving the real gray level of the object under the condition of double energy transmission,low energy pre-scattering and backscattering.And then we evaluated the model.The results show that the real gray scale of the object is more accurate.Finally,in view of the difficulty in liquid recognition with X-ray safety detection technology,the combination of X-ray detection and electronic nose odor recognition is proposed to determine the liquid properties in the container.We combined the image contour,image gray level,electronic nose response and other detection information to build the model of multi-information fusion detection method and built the hardware and software system of multi-information fusion detection technology.Then we use different pattern recognition methods to process the experimental data.We mainly studied the pattern recognition method of neural network and then established an effective BP network model.This paper made a new attempt to study multi-information and high-performance security equipment and realized the fusion of multiple detection information.It provides an effective method and theoretical basis for the research and development of new security equipment.It is proved from the theory,experiment and practice that multi-information fusion security detection is an effective scheme in the postal detection system.