基于神经网络技术的遥感水深反演模型研究

利用Landsat7 ETM+遥感影像反射率和实测水深值之间的相关性,选取了相关性较好的ETM1、ETM2、ETM3、ETM4、ETM3/ETM2等5个水深反演因子,建立了BP神经网络水深反演模型。为充分体现BP神经网络模型的优越性,利用SPSS软件建立了单波段、波段比值、多波段三种不同的线性回归模型。通过对比发现,具有很好的自适应能力和非线性映射能力的BP神经网络模型在处理遥感水深反演问题上比传统的线性模型效果更好。

Water Depth Remote Sensing Retrieval Model Based onArtificial Neural Network Techniques

A momentum and self-adaptive BP neural network model was constructed to retrieve the water depthinformation using the relationship between reflectance derived from Landsat ETM+satellite data and water depthinformation.The BP neural network model was established by these five factors:ETM1,ETM2,ETM3,ETM4,ETM3 / ETM2.In order to fully reflect the superiority of the BP neural network model,three linear regressionmodels including single-band model,dual-band ratio model and multi-band model were established using theSPSS software.Through comparison,it was found that the momentum and self-adaptive BP neural network modelhad a better performance than the traditional linear regression model in the retrieval of water depth.