摘要
传统的打印机感光鼓损伤印迹识别系统识别时间长,工作过程稳定性差,为了解决上述问题,基于深度学习设计了一种新的打印机感光鼓损伤印迹识别系统。在传统卷积神经网络基础上增加训练模型,设计了一种新的卷积神经网络,将该网络引入到系统硬件中,构建出系统硬件总结构,包括输入层、卷积层、下采样层、特征提出层和输出层五层。根据硬件设计软件程序,共分为信息预处理、感光鼓损伤印迹定位、感光鼓损伤印迹分割、感光鼓损伤印迹识别四步。为检测识别系统的工作效果,与传统识别系统进行实验对比,结果表明,基于深度学习设计的识别系统能够有效缩短感光鼓损伤印迹识别时间,工作过程稳定性高。
Since the traditional printer photosensitive drum damage imprint recognition system takes a long time for the recognition and the working stability is poor,a new printer photosensitive drum damage imprint recognition system is designed based on the deep learning. In the system,a new convolutional neural network is designed by adding a training model to the traditional convolutional neural network. The new network is introduced into the system hardware to construct the overall system hardware structure,which include five layers,named input layer,convolutional layer,downsampling layer,feature extraction layer and output layer. According to the hardware,the software program is designed,which is divided into four steps,named information pretreatment,location of photosensitive drum damage imprint,segmentation of photosensitive drum damage imprint and identification of photosensitive drum damage imprint. Contrastive experiments on the proposed recognition system and traditional recognition system were carried out to test the operation effects. The results show that the proposed recognition system based on the deep learning can effectively shorten the recognition time of photosensitive drum damage imprint,and is of high working stability.
作者
李志荣
LI Zhirong(Department of Document Examination Technology,Criminal Investigation Police University of China,Shenyang 110035,China)
出处
《现代电子技术》
北大核心
2020年第11期163-166,171,共5页
Modern Electronics Technique
关键词
感光鼓损伤印迹识别系统
深度学习
卷积神经网络
实验参数设定
识别时间对比
稳定性对比
photosensitive drum damage imprint recognition system
deep learning
convolutional neural network
experimental parameter setting
recognition time contrast
stability contrast
