论文部分内容阅读
提出了基于多波束栅格图像和改进神经网络的底质分类方法,研究了多波束栅格回波强度的提取方式和改进的反向传播(BP)神经网络.论述了波束脚印包络内以采样数量进行等角度栅格分配获取回波强度所在位置序列值,并在传统BP神经网络基础上附加动量因子和自适应学习率,同时为激活函数添加斜率和偏置可随误差信号进行修正.改进的BP神经网络不仅可以提高神经元的自适应能力,而且可以明显加快算法的收敛速度.利用提出的方法进行底质分类,实验结果表明,提出的方法显著提高了海底底质分类的分辨率和精度.
This paper presents a method for the classification of ground objects based on multi-beam grid images and improved neural network, and studies the extraction methods of multi-beam grid echo intensities and the improved back propagation (BP) neural network. The number of samples is equi-angularly distributed to obtain the position value of the echo intensity. Based on the traditional BP neural network, a momentum factor and an adaptive learning rate are added. Meanwhile, the slope and offset for the activation function can be corrected with the error signal. The improved BP neural network can not only improve the adaptive ability of neurons, but also speed up the convergence of the algorithm significantly.Using the proposed method to classify the sediments, the experimental results show that the proposed method significantly improves the resolution of the submarine sediment classification And precision.