基于多层编码器的SAR目标及阴影联合特征提取算法

针对合成孔径雷达(SAR)图像目标识别问题，提出一种基于多层自动编码器的特征提取算法。该方法利用随机神经网络受限波尔兹曼机学习建模环境概率分布的能力，通过组建更具函数表达能力的多层神经网络，提取描述目标及其阴影轮廓形状的综合特征。利用两种分类模型实现目标自动识别。基于MSTAR 数据的仿真实验结果验证了算法的有效性。 

Shared Representation of SAR Target and Shadow Based on Multilayer Auto-encoder

Automatic Target Recognition (ATR) of Synthetic Aperture Radar (SAR) image is investigated. A SAR feature extraction algorithm based on multilayer auto-encoder is proposed. The method makes use of a probabilistic neural network, Restricted Boltzmann Machine (RBM) modeling probability distribution of environment. Through the formation of more expressive multilayer neural network, the deep learning model learns shared representation of the target and its shadow outline reflecting the target shape characteristics. Targets are classified automatically through two recognition models. The experiment results based on the MSTAR verify the effectiveness of proposed algorithm.