Study On Seal Imprint Time Verification Based On Neural Network

With increasing various civil economic criminal cases using the forging documents imprint,the documents forming time identification is becoming a problem needing to be solved to the document examination expert.The seal and print as a main evidence of reviewing documents' authenticity and effectiveness,more and more dispute about the seal imprint verification not only the identify problem but also focus on the identification between the seal imprint time and the documents forming time.Based on the seal and print variability characteristics to identify the seal imprint time is an effective way to solve this kind of problems.With the rapid development of the computer technology and the widely used of the ANN on the data mining,this thesis proposes a data mining method based on BP neural network to exploit the experts' empirical data,the method finally realized the computer aided identification on the seal imprint time.The main research contents of this thesis are as follows:Firstly,this thesis presents the seal characteristic value index system based on the expert experience.Through the seal and print variability characteristics recognized by the document examination expert,this thesis analyses the seal and print variability characteristics that affects the identification of the seal imprint time,and summarizes a more scientific and reasonable characteristic value index system based on the statistics analysis about the seal characteristic.Secondly,this thesis extracts and quantizes the characteristic parameters using computer aided recognition based on the seal characteristic value index system. Meanwhile the qualitative characteristic data is converted into the quantitative characteristic data,and it provides reasonable and reliable data supports for the BP neural network application.Based on analysis the theory of using BP neural network to identify the seal imprint time,this thesis establishes the recognition model using the three-tier feed forward neural networks,and discusses in detail the neural network topology structure, the numbers of the neuron in hidden layer,the sample data selection and preprocessing, the initial parameters determination,the activation function selection and so on.The BP learning algorithm are realized by used C# as development language,takes some types of seal in some companies as an example,establishes the seal imprint time verification model based on BP learning algorithm.Through the simulation experiment on the sample data and test data,it is proved that the model has high recognition accuracy and a better generalization ability.Then using the recognition model on the seal imprint time verification field,realizes the seal imprint time verification scientific and automatic, meanwhile the research shows that the effectiveness and practicality of using the BP neural network on seal imprint time verification.