摘要
主要利用人工神经网络的理论知识研究在图像识别中的应用为目的,研究图像识别中图像分割的技术,同时详细分析了多层前馈神经网络的描述及BP算法工作过程。介绍隐层的选择及隐层神经元数选择的一些经验方法。针对BP算法存在的问题,提出加可变动量因子的BP算法,通过对网络训练过程参数调整以及增加可变动量因子等方面进行优化改进,实验证明加快了训练速度,改善了BP网络的学习效果。
By using the knowledge of artificial neural network,the application and image segmentation im image recognition are researched,the multi- layer feedforward neural rietwork and process of BP algorithm are analysed in a detail. Sorne methods of hidden tier and hidden tier neuron chosen are introduced. Aiming at Problems of BP,variable factor BP algorithm is proposed. By adjusting network training process parameter and optimizing variable factor, the experiment proves that training speed is incteased,study effect of BP network is improved.
出处
《现代电子技术》
2008年第8期127-130,134,共5页
Modern Electronics Technique
关键词
人工神经网络
BP神经网络
图像分割
artificial neural network
BP neural network
image segmentation
