Subpixel Target Detection Based on Multi-Layer Neural Networks

The first objective of this paper is to develop a gradient descent algorithm mathematically for solving the stochastic optimization problem of a multi-layer neural network numerically using a variable step-size that is determined by the algorithm automatically. Artificial neural networks have difficulty in hyperspectral imaging for target detection of rare targets. The difficulty is caused by the scarcity of available target pixels in the image for training and testing the neural networks. The second objective of this paper is to analyze the convergence of the proposed algorithm and then apply the algorithm to subpixel target detection in hyperspectral imaging using a real hyperspectral image. Additional target pixels are generated using synthetic signature modeling. The decision boundary for the multi-layer neural network is nonlinear for the transformed pixels obtained through whitening transformation. Experimental results show that the proposed algorithm converges and can perform well in subpixel target detection.