摘要
基于深度卷积神经网络算法实现网络图像超分辨率重建技术,为满足图像的超分辨率精度检测和构建需求,通过构建图像融合技术来实现图像重建架构,形成以机器人视觉系统数据为主体的控制模块,实现对网络图像超分辨率的图像融合分析的目标,完成深度卷积神经网络图像重建。在深度卷积神经网络图像的构建过程中,注意神经网络输出数据决策方案和图像的自适应预置模块设计,分析深度卷积神经网络的各层节点数,平衡图像分辨率数据深度卷积过程中的信息损失量,提升图像分辨率数据的重建精度。
The network image super-resolution reconstruction technology is realized based on the deep convolution neural network algorithm.In order to meet the needs of image super-resolution accuracy detection and construction,the image reconstruction framework is realized by constructing the image fusion technology,forming a control module with the robot vision system data as the main body,realizing the goal of image fusion analysis of network image super-resolution,and completing the deep convolution neural network image reconstruction.In the construction process of deep convolution neural network image,pay attention to the decision-making scheme of neural network output data and the design of adaptive preset module of image,analyze the number of nodes in each layer of deep convolution neural network,balance the amount of information loss in the process of deep convolution of image resolution data,and improve the reconstruction accuracy of image resolution data.
作者
陈欣
王凌
朱佳佳
刘子寒
沈力
Chen Xin;Wang Ling;Zhu Jiajia;Liu Zihan;Shen Li(State Grid Jiangsu Electric Power Co,LTD Information&Telecommunication Branch,Nanjing 210024,China)
出处
《单片机与嵌入式系统应用》
2023年第1期7-10,共4页
Microcontrollers & Embedded Systems
关键词
卷积神经网络
图像融合
数据决策
分辨率
convolutional neural network
image fusion
data decision
resolving power
