The advancement of modem technology has led to the rapid development of image information recognition systems.In order to improve the recognition accuracy of information in digital images and ensure the security of user related data information,the experiment proposes to apply the Areal feature recommendation network to digital image information recognition.In the process,the image information is preprocessed by using depth learning,and then the final recognition of image information is completed by using the Areal feature recommendation neural network.The results show that the research method is tested on MNIST data set and Cifar da-ta set.When the iteration reaches the 20th and 16th times,the research method has stable Loss function values of 0.015 1 and 0.0163.When the accuracy of all models is 0.800,the recall rates of RNN,TensorFlow CNN,and research methods obtained are 0.722,0.784,and 0.902,respectively.When studying methods for image recognition simultaneously,there was no overlap phenomenon.The above results indicate that the research method has a high accuracy in image recognition,which can provide a certain reference for the performance improvement of subsequent image recognition systems and has high application value.
关键词
数字图像/信息识别/区域特征推荐神经网络/深度学习
Key words
digital images/information identification/areal feature recommendation neural network/deep learning
维普
万方数据