Underwater acoustic image segmentation using neural deformable template

This paper describes a neural network architecture that has been developed to perform underwater acoustic image segmentation based on deformable template techniques. The template shift is guided by external energy from the image under consideration, which attracts them towards the target characteristics and by internal energy, which tries to maintain the smoothness of the contour curve. The resulting deformable templates are used to train and test the neural network, which is able to distinguish interest objects from other kinds of objects. We have used the neural deformable template to segment underwater acoustic images. The results show that this algorithm is efficient, precise and very immune to image deformation and noise when compared to results obtained from several other deformable template-based methods.