| 一种基于神经网络的遥感图像压缩编码 |
| |
| 引用本文： | 张煜东 吴乐南 王水花,孙晓燕.一种基于神经网络的遥感图像压缩编码[J].南京信息工程大学学报,2009,1(1):82-88. |
| |
| 作者姓名： | 张煜东 吴乐南 王水花 孙晓燕 |
| |
| 作者单位： | 1. 东南大学信息科学与工程学院,南京,210096
2. 奥兰科技大学计算机科学学院,信号图像语音识别实验室,奥兰,阿尔及利亚 |
| |
| 基金项目： | 国家863计划，国家自然科学基金，高等学校科技创新工程重大项目培育资金，江苏省自然科学基金，东南大学优秀博士学位论文基金，河海大学青年科技基金 |
| |
| 摘 要： | 为了更好地在恒定压缩率条件下，实现卫星遥感图像的压缩，提出1种将像素分为3类的编码方案．每类采用基于Levenberg-Marquardt算法的双层神经网络进行预测，和基于偏置学习规则的竞争神经网络实现量化．对长城、珠峰、香港等地遥感图像的压缩试验证实了算法的有效性．同时还表明该算法压缩图像的MSE高于距离-权值方法与最小均方误差法，且时间复杂度为O（n）．
|
| 关 键 词： | 神经网络 遥感 图像编码 |
| 收稿时间： | 2009/5/29 0:00:00 |
A neural network based compression coding for remote sensing images |
| |
| Affiliation: | ZHANG Yudong WU Lenan WANG Shuihua Neggaz Nabil SUN Xiaoyan( 1 School of Information Science ＆ Engineering,Southeast University,Nanjing 210096 2.Lab. of Signal Image Parole, Dept. of Computer Science, University of Science and Technology of Oran, Oran, Algeria) |
| |
| Abstract: | To compress remote sensing images at a fixed ratio, a novel coding scheme is proposed in which pixels are divided into three types. And different prediction and quantification are adopted for each type, in which a twolayer neural network based on Levenberg-Marquardt algorithm is chosen for prediction model, and a competitive neural network is selected for quantification. Compression experiments on remote sensing images of Great Wall, Mount Everest,and Hong Kong indicate the validity and effectiveness of our algorithm. Moreover, the MSE of the proposed method is higher than distance-weight method and least mean square error method, and the computation complexity of our method is O（n）. |
| |
| Keywords: | neural network remote sensing image coding |
| 本文献已被 维普 万方数据 等数据库收录！ |
| 点击此处可从《南京信息工程大学学报》浏览原始摘要信息 |
|
点击此处可从《南京信息工程大学学报》下载全文 |
|
