| 基于小波变换和混合神经网络的图像压缩算法 |
| |
| 引用本文： | 李万臣,王炼. 基于小波变换和混合神经网络的图像压缩算法[J]. 应用科技, 2006, 33(1): 29-31 |
| |
| 作者姓名： | 李万臣 王炼 |
| |
| 作者单位： | 哈尔滨工程大学,信息与通信工程学院,黑龙江,哈尔滨,150001 |
| |
| 摘 要： | 提出了一种将空间方向小波零树编码与混合神经网络相结合，新的多尺度系数矢量量化策略．该算法在对图像进行多级小波变换后，利用3个方向上各自小波系数之间的相关性，构造符合图像特征的跨频带矢量，依据矢量能量和零树矢量的思想进行矢量分类，分别利用主元分析和自组织特征映射神经网络对3个方向的多尺度系数矢量进行基于视觉的加权矢量量化压缩编码．仿真实验结果表明该算法是合理可行的．
|
| 关 键 词： | 小波变换 图像压缩 矢量量化 神经网络 |
| 文章编号： | 1009-671X（2006）01-0029-03 |
| 收稿时间： | 2004-12-20 |
| 修稿时间： | 2004-12-20 |
An image compression algorithm based on wavelet transformation and mixed neural network |
| |
| LI Wan-chen,WANG Lian. An image compression algorithm based on wavelet transformation and mixed neural network[J]. Applied Science and Technology, 2006, 33(1): 29-31 |
| |
| Authors: | LI Wan-chen WANG Lian |
| |
| Abstract: | A new strategy of multi-dimensional coefficient vector quantization was proposed, in conjunction with wavelet zerotree encoding and mixed neural network. This algorithm constructs cross-band vector which conforms to the image characteristics by using the correlations of wavelet coefficients in 3-directions after the muhi-level wavelet transformation to images. The vectors are classified in accordance to vector energy and zerotree vectors. Finally, the multi-dimensional coefficient vectors are compressively coded with respect to vision by using PCA/SOFM neural network and HVS. The simulation shows the feasibility of this algorithm. |
| |
| Keywords: | wavelet transformation image compression vector quantization neural network |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
