3D Millimeter Wave Image Prohibited Item Recognition Based On Deep Learning

After the "9-11" terrorist incident,the security inspections of airports around the world have become increasingly strict,and with the various dangerous goods,the security situation of the airport is extremely severe.At the same time,with the development of the economy,the flow of people at the airport has exploded,and the security of the airport is facing tremendous pressure and challenges.In order to improve the efficiency and quality of airport security inspection,improve the security satisfaction of passengers,and reduce the workload of airport security personnel,the article conducted a study on the safety inspection of human body belongings in airport security inspection.The main work of the article is as follows:The high-efficiency identification of prohibited items in 3D millimeter-wave scanning images in airport human security inspections has been studied: millimeter-wave human body imaging technology is an advanced technology in the global security field,and has been used for passengers' personal security inspections in airports in the United States,Britain and other countries.However,the efficient identification of prohibited objects in millimeter-wave images is still a difficult problem to be solved,which greatly limits the application of millimeter-wave imaging technology in airport passenger screening.The article applies deep learning technology to 3D human body millimeter-wave images,and automatically discriminates whether there are prohibited items in the millimeter-wave image and where the prohibited items are hidden in the body.The Convolutional Neural Network(CNN)is used to extract the features of the millimeter-wave images of multiple views of the human body,and then the multi-view image features are merged through the Long Short-term Memory(LSTM)in the feature fusion process.The attention mechanism was introduced to further enhance the feature fusion effect.Finally,the multi-channel Sigmoid classifier was used to obtain the probability values of the prohibited items in the 17 parts of the human body.In the experiment,the loss error on the test set was 0.03.Of the 100 test samples,80 contained prohibited items,and the remaining 20 had no prohibited items.The article algorithm identified 76 samples with prohibited items and 24 no items.For the identification of prohibited items,the false positive rate is 0%,and the missed detection rate is 5%,indicating the effectiveness of the method.Aiming at the prohibited item identification algorithm proposed in the article,a prototype application system for prohibited items identification based on Web application is designed.The web client transmits the image data to be recognized through the millimeter wave imaging system,and the server predicts the image data through the trained prohibited item identification model.And return the processed prediction result to the client to display the prohibited item identification result.Through the system to automatically identify the prohibited items in the millimeter-wave image of the human body,while ensuring safety,the millimeter-wave scanned image of the passenger is also prevented from being exposed to other people,thereby protecting the personal privacy of the passenger.