Regression Neural Network for Soil Moisture Estimation

Abstract

Over the past few years image categorization using deep learning became very popular, because it can handle large databases and has shown good recognition results. This paper presents the complex valued convolutional network (CV-CNN) for Synthetic Aperture Radar (SAR) soil moisture estimation. The CV-CNN consist in general of a real or complex valued input layer, output layer and one or more hidden layers. Hidden layers represent any combination of convolutional layers, pooling layers, activation functions, and are fully defined within complex valued domain. This paper proposes a deep Convolutional Neural Network (CNN) for soil moisture parameter estimation. A 9 layer convolutional neural network was used, consisting of convolutional, pooling, dropout, fully connected and regression layers. We used 1000 ground measurements for each SAR acquisition using L-band fully polarimetric SAR data. The experimental results were verified using 100 ground points.