基于BP神经网络的DEM粗差探测法分析
Analysis of DEM Gross Error Detection Method based on BP neural network
陈百了 1周访滨1
作者信息
摘要
阐述一种以BP神经网络为基础的DEM粗差探测方法,为了减少该算法的误判率,加入怀疑率这一参数,这样某一点的粗差判断可以根据多个区域的检测结果来进行.为了确定怀疑率的阈值,分别采用冒泡法、均值法、中位数法.其中中位数法的实验效果更优.以中位数确定的怀疑率阈值进行实验,检测率达到94.1％,误判率为31.5％.
Abstract
This paper expounds a DEM gross error detection method based on BP neural network.In order to reduce the misjudgment rate of the algorithm,the parameter of suspicion rate is added,so that the gross error judgment of a certain point can be based on the detection results of multiple regions.To determine the threshold of suspicion rate,bubble method,mean method,and median method were used respectively.The median method has better experimental results.The experiment was conducted with a suspicion rate threshold determined by the median,and the detection rate reached 94.1％,with a misjudgment rate of 31.5％.
关键词
深度学习/BP神经网络/数字高程模型/粗差探测Key words
deep learning/BP neural network/digital elevation model/gross error detection引用本文复制引用
基金项目
长沙理工大学研究生科研创新项目(SJCX2021121)
出版年
2024
NETL
NSTL
万方数据