Research On Fine-grained Metal Detection Based On Frequency Domain

With the development of science and technology,the comprehensive strength of all countries in the world has been significantly improved.And people’s living standards have also risen significantly too However,there are still a few wars in some areas.Some countries even have political turmoil and terrorism.Nowadays combating terrorism has become an important issue for the international community.In view of this,our country has stepped up security and prevention work in public places from various aspects to counter the threat of terrorist attacks.Setting up specialized security checks in densely populated places such as the subway,shopping malls,railway stations and concerts is the most effective and most direct way to guard against attacks.The main purpose of security is to inspect prohibited items,such as some common metal products(knives,scissors,guns,nunchaku,etc.),flammable and explosive materials,drug and so on.One of the most important part of the detection is metal detection.Currently metal detection and identification methods are mainly divided into two categories: metal detection based on electromagnetic induction and metal detection based on electromagnetic wave.Electromagnetic induction based detectors need to be very close to the target required to achieve accurate detection.Electromagnetic wave based detectors such as X-ray detectors and walk-through metal detector.It has the following drawbacks: very high cost of implementation and difficult to transport and move.Because the products usually very bulky,so they only install in entrance of public places.What’s more,X-ray can bring some serious harm to the human body.And the use of personnel also need to have certain technical requirements.Based on the above problems,metal detection equipment can not achieve large-scale deployment.Today more and more smart devices are used.Many public places have full coverage of wireless signals,giving us a lot of convenience.If we can detect metal via wireless signals,we can deploy them on a large scale at relatively low cost.Metal in the dielectric constant and other physical properties and non-metallic substances exist some differences.We use the shielding effect of metal in electromagnetic waves to distinguish between metal and nonmetal.In short,metal has a strong reflection of electromagnetic waves.Only a small amount of electromagnetic waves can penetrate,more often can not penetrate.The reflecting ability of non-metallic materials is relatively strong.And most signals can penetrate non-metallic for transmission.So we use the characteristics of electromagnetic waves to detect metal.In order to solve the above problems,we presents a method that can identify metal by using existing wireless networks and devices in a specific indoor environment.The main work of this paper can be divided into three parts.First,metal identification research is conducted with the help of Channel State Information(CSI).The CSI is channel information at the physical layer of the wireless network and contains a plurality of subcarrier amplitude and phase information.In contrast to the Received Signal Strength Indication(RSSI),the channel state information is more granular.So CSI is more useful for our metal detection identification.Through the study of CSI,we can observe the impact of metal on the propagation of wireless signals.Second,we designs and realizes the metal detection system,including hardware design,software design,scene design and system architecture design.The system experiments in the real life scene,and uses the existing commercial wireless network equipment.Third,we analysis and process the data of channel status information.The filter is used to preprocess the data of channel state information in order to minimize the noise due to environmental factors.Then we extract feature and analysis.Finally,we use the sparse classifier to improve the accuracy of the system for metal detection.The related research and experiments show that the metal detection system in this paper has achieved the expected results,but still has room for further improvement.We conducted experiments to evaluate its performance in two typical scenarios.The results demonstrate that our method can achieve an average recognition accuracy of 99.3% in Chamber scenario and average accuracy of 94.5% in the Lab experimental scenario.The main innovation of this paper lies in the first establishment of a system for building metal detection through existing wireless network equipment and related technologies.The system is simple,easy to deploy,low cost,has some practical significance,and also has a very broad application prospects.