目 录
摘 要................................................................................................................................................1
Abstract............................................................................................................................................2
第一章 绪 论...................................................................................................................................3
1.1 引 言..................................................................................................................................................................3
1.2 手写体数字识别的目的意义...........................................................................................................................3
1.3 手写体数字识别的现状...................................................................................................................................3
1.4 手写体数字识别的发展...................................................................................................................................4
1.5 本文研究的主要内容.......................................................................................................................................5
第二章 人工神经网络....................................................................................................................6
2.1 神经网络概述...................................................................................................................................................6
2.1.1 人工神经网络发展经历................................................................................................................6
2.1.2 神经网络的基本原理....................................................................................................................7
2.2 BP 神经网络.....................................................................................................................................................9
2.2.1 BP 神经网络的网络模型..............................................................................................................9
2.2.2 BP 神经网络的基本原理..............................................................................................................9
2.2.3 BP 学习算法的流程....................................................................................................................12
2.3 神经网络在模式识别问题上的优势.............................................................................................................13
第三章 手写体数字识别常规预处理和特征提取方法..............................................................14
3.1 手写体数字识别的识别过程.........................................................................................................................14
3.2 手写体数字识别的常规预处理方法.............................................................................................................15
3.2.1 图像的灰度化..............................................................................................................................15
3.2.2 二值化..........................................................................................................................................16
3.2.3 平滑去噪......................................................................................................................................17
3.2.4 书写的倾斜矫正..........................................................................................................................17
3.2.5 数字图像分割..............................................................................................................................18
3.2.6 图像的归一化处理......................................................................................................................18
3.2.7 图像的细化..................................................................................................................................19
3.3 特征提取.........................................................................................................................................................19
第四章 基于 BP 算法的手写体数字识别的实现........................................................................21
4.1 基于 BP 神经网络手写体数字识别的过程..................................................................................................21
4.2 样本准备.........................................................................................................................................................21
4.2.1 灰度化..........................................................................................................................................22
4.2.2 二值化..........................................................................................................................................22
4.2.3 平滑去噪......................................................................................................................................22
4.2.4 数字图像归一化..........................................................................................................................23
4.2.5 细化..............................................................................................................................................23
4.3 特征提取.........................................................................................................................................................23
4.4 神经网络结构参数的选择与确定.................................................................................................................24