| 基于卷积神经网络的模式分类器 |
| |
| 引用本文： | 李葆青.基于卷积神经网络的模式分类器[J].大连大学学报,2003,24(2):19-23. |
| |
| 作者姓名： | 李葆青 |
| |
| 作者单位： | 六盘水师范高等专科学校,贵州 六盘水 553004 |
| |
| 摘 要： | 研究了使用卷积神经网络((Convolutional Neural Networks,CNNS)构造模式分类器,并用于文本/图像分割和文本检测的可能性.CNNs可以避免显式(直接)特征取样.更为重要的是,CNNs能直接运作于灰度图像,使其应用变得直截了当.对诸如卷积核尺度、网络收敛速度等具体算法实现问题进行了讨论,并给出CNNs在汉字文本/图像分割和文本检测方面的各种实验结果.
|
| 关 键 词： | 模式分类 卷积神经网络 人工智能 |
| 文章编号： | 1008-2395(2003)02-0019-05 |
| 修稿时间： | 2002年12月26 |
Building pattern classifiers with convolutional neural networks |
| |
| LI Bao-qing.Building pattern classifiers with convolutional neural networks[J].Journal of Dalian University,2003,24(2):19-23. |
| |
| Authors: | LI Bao-qing |
| |
| Abstract: | This paper studies the possibility of buiding pattern clssifiers for text/picture segmentation and text detection problems with convolutional neural networks (CNNs). Using CNN, explicit feature extraction can be avoided. More importantly, CNN can directly operate on grey level images, making its application straightforward. The issues such as kernel size, convergence speed, etc. are discussed. The results of the experiments on Chinese text/ picture segmentation and text detection are presented. |
| |
| Keywords: | pattern classification convolutional neural networks artificial intelligence |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
