Design of Neural Network-based Intelligent Extraction Method for Key Electronic Information

Abstract

At present of rapid development of information technology, it is of great practical significance to extract and analyze electronic information. This paper combines ORB features and an improved convolutional neural network, establishes ORB feature extraction by fast key point extraction and rBRIEF, and also improves the data layer, convolutional layer and loss layer of convolutional neural network to construct the information extraction network model, and images the electronic information to obtain more key information. Taking the forest resources of land M as an example for application research, remote sensing data are collected, and the research is carried out in terms of the dynamic changes of different levels of forest resource information extraction. The results can be obtained that the forests in the middle and eastern parts of Land M have a low degree of depression, with an average diameter at breast height (DBH) of less than 18 cm and a stock volume of less than 100 m3/ha, while in the western part, the degree of depression and the average diameter at breast height (DBH) are higher, and the density of the forest stands in each region is concentrated at 500-1000 plants/ha. The overall forest cover of the land showed an increasing trend from 2016 to 2020. In general, from 2019 to 2020, the area of broadleaf forests, coniferous forests, and non-forested land increased, while the area of bamboo forests decreased. Broadleaf and coniferous forests saw an increase of 14.25% and 3.11% respectively. The area of bamboo forest decreased by 15.42%. The effectiveness of the constructed method for intelligent extraction of key electronic information was verified through application analysis.

Keywords

• orb features
• convolutional neural network
• information extraction
• forest resources
• 94a08