Research On The Method Of Airport Passenger Security Witness Integration In Complex Lighting Environment

With the continuous development of e-commerce,finance,security protection and traffic,face and identity recognition technology has played a huge role in more and more fields.Traditional airport security inspections have certain limitations.Lack of efficient and accurate identification methods for customs clearance holders holding false documents and other people’s documents,and there are great security risks.The widespread use of face and identity card integration technology makes identity Verification becomes efficient and convenient,which not only improves the accuracy of identification,but also guarantees the security of the airport.Under the ideal airport security environment,the face recognition rate is high.At present,a small number of airports in Xinjiang are affected by complex lighting.Factors such as direct exposure in the morning and backlight at night have led to the loss of some face image information collected on the scene,which has affected the recognition rate.At the same time,most of the ID cards are taken ten years ago or longer,and some facial features will change with time.If the ID card image is directly matched with the face image collected on the scene,the recognition rate will be reduced.The thesis relies on the key research and development project of the Sichuan Province Science and Technology Plan "Research on the key technologies of the passenger airport self-service security check system",through the adaptive pre-processing of face images collected on the scene to reduce the impact of complex lighting;The feature of extracting the face image from the influence of time is used to solve the problem of matching verification of the face ID image across ages.The research contents and methods of this thesis are as follows:In the research of face image illumination preprocessing methods,the principles of common methods are analyzed and examples are verified.By comparing the advantages and disadvantages of each method,an adaptive image segmentation method based on the combination of Otsu and K-means is proposed.First,adaptively segment regions of the face image that are affected by different illumination,and then perform gamma correction on each region to reduce the impact of complex illumination.Using data to verify and analyze related methods,it is concluded that the method in this thesis is better than other methods under different light effects.In terms of research on feature extraction methods of face images that are not affected by time,this thesis selects geometric features as stable features of face images.Analyze the principle of common feature point location methods and verify the examples.By comparing the advantages and disadvantages of each method,a feature point location method based on the fusion of active appearance model(AAM)and gray co-occurrence matrix(GLCM)features is proposed.The shape and texture of the sample are modeled first,and then the gray correlation information near the sample feature points is counted.Using data to verify and analyze the relevant methods,it is concluded that the method in this thesis can more accurately locate the geometric feature points of the face image of the ID card than the other methods.Based on the above research,the development of the face authentication system is completed to realize the recognition and verification of the face and ID card images under complex lighting conditions.