A New Kernel-Based Neural Network with Application in Sea Target Detection

Sea target detection is an important goal for military purposes and navigation. Environmental noise and sea clutter are two major problems in sea target detection. A new novel Neural Network in Kernel Space (NNKS) is presented for classification of the target and non-target in color images, and it is followed with a new kernel neuron sea target detection system (KNNS) which is efficient especially in the critical color images with clutter, noise and complex background that relative methods are not effective. The neural network is trained with the color characteristics of the target and non-target. In this way, color features of the HSV color space are applied to train the NNKS neural network. The superiority of the new method to the relative previous works and other color target detection based methods is conspicuous.