| 一种新的基于Kohonen网络的图像矢量量化 |
| |
| 引用本文： | 张传生,丁承民,刘贵忠.一种新的基于Kohonen网络的图像矢量量化[J].系统工程与电子技术,1997(3). |
| |
| 作者姓名： | 张传生 丁承民 刘贵忠 |
| |
| 作者单位： | 西安交通大学电信学院 710049 |
| |
| 摘 要： | 矢量量化是一种重要的数据压缩方法.本文利用Kohonen自组织映射神经网络进行矢量量化,首次较为详细地讨论了具体实现的步骤与细节,并在此基础上为改善边缘质量,提出一种基于Laplas算子检测边缘弓引导神经网络训练的方法,并通过实验证明其效果是明显的.
|
| 关 键 词： | 网络 矢量量化编码 数据压缩 神经网络 |
A New Approach for Vector Quantization Based on Kohonen Neural Network |
| |
| Zhang Chuansheng,Ding Chengmin and Liu Guizhong School of Electronics Communication,Xi'an Jiaotong University.A New Approach for Vector Quantization Based on Kohonen Neural Network[J].System Engineering and Electronics,1997(3). |
| |
| Authors: | Zhang Chuansheng Ding Chengmin and Liu Guizhong School of Electronics Communication Xi'an Jiaotong University |
| |
| Institution: | Zhang Chuansheng,Ding Chengmin and Liu Guizhong School of Electronics Communication,Xi'an Jiaotong University,710049 |
| |
| Abstract: | This paper extensively discusses the procedure for vector quantization based on Ko-honen self organizing feature map and uses a relative new version of frequency sensitive technology. The most important feature of it is to use a laplas operator to detect the image edge and induct the neural network to learn more information about the high frequency area in order to diminish block effects and therefore reconstruct the image with higher fidelity. The experiment results prove the effectiveness of this new approch. |
| |
| Keywords: | Vector quantization Laplas operator Neural network LFKVQ |
| 本文献已被 CNKI 等数据库收录！ |
