Forgery Numeral Handwriting Detection based on Convolutional Neural Network

Handwriting forgery detection is one of the hot spots in forensic science, and economic cases of handwritten forged numbers are increasing. At the same time, forgery identification of documents is an important evidence in criminal proceedings. For the problem of tedious and low degree of automation of manual document inspection, put forward a method for handwritten forged numeral detection based on convolutional neural networks. The experiment convened 50 volunteers and collected image samples of 6 types of forged and normal handwritings with 50 different brand pens, and established a total of more than 7,200 data sample sets. Perform parameter optimization on the basis of Alexnet, while training and testing the data set. The experimental results show that the highest accuracy rate of the six types of handwritten forged numeral recognition is 97.84%, and the average accuracy is 95.35%, which is better than the SVM feature classifier. It provides a new method for forgery handwriting identification.