基于机器学习SAR图像噪声抑制技术研究 |
| |
引用本文: | 徐东,张晓雯.基于机器学习SAR图像噪声抑制技术研究[J].人天科学研究,2013(10):163-165. |
| |
作者姓名: | 徐东 张晓雯 |
| |
作者单位: | 海军大连舰艇学院基础部,辽宁大连116018 |
| |
摘 要: | 为提高含噪声SAR(合成孔径雷达)图像的目标识别能力,提出基于机器学习的图像噪声抑制技术的研究思路。该思路通过问题估计,为舍噪声图像获得清晰的输出景物,从而生成一个与含噪声图像相符合的增强的合成景物世界。该技术以马尔可夫网络为体系,根据图像与景物、景物与景物之间的联系来确定网络上的信息传递规则,从大量的训练事例中学习并获得网络参数,利用贝叶斯方法为原图像找到理想的后验概率,生成一个清晰的超分辨率结果图。
|
关 键 词: | SAR图像 马尔可夫网络 贝叶斯 噪声抑制 |
Research on Machine Learning Based SAR Image Noise Blanking Technology |
| |
Abstract: | This paper suggested an idea of machine leaning based image noise blanking technology for improving the ability target recognition of SAR noise image, the idea was to obtain a clear scene for the noise image , through the problems- es- timating , It could generate a compatible with noise image and enhanced synthesis of scene image ~ The technology mod- eled that world with a Markov network which learned from a number of training examples and then got the network pa- rameters according to image and scene, scene and scene to determine the link between the messages on the network rules. Bayesian method allowed it efficiently find a fitting posterior probability for the scene, generating a clear high resolution images. |
| |
Keywords: | SAR Image Markov Network Bayesian Noise Blanking Scene |
本文献已被 维普 等数据库收录! |
|