采用形态学分量分析的红外云图插值  

Infrared nephogram interpolation based on morphological component analysis

作　　者：周峰[1] 金炜[1] 符冉迪[1] 

机构地区：[1]宁波大学信息科学与工程学院,浙江宁波315211

出　　处：《宁波大学学报（理工版）》2017年第1期110-115,共6页Journal of Ningbo University：Natural Science and Engineering Edition

基　　金：国家自然科学基金(61271399;61471212);宁波市科技创新团队研究计划(2011B81002);浙江省自然科学基金(LY16F010001);宁波市自然科学基金(2016A610091);浙江省信息与通信工程重中之重学科项目(XKXL1306;XKXL1425)

摘　　要：提出一种基于形态学分量分析的红外云图插值重建方法.利用平稳小波变换和轮廓波变换在表示平滑分量和纹理分量时各自的优势,将得到的低分辨率红外云图通过形态学分量分析方法分解为平滑分量和纹理分量;并针对小波插值使低频能量损失问题,采用先插值后变换的方法改进平稳小波插值效果,分别用平稳小波变换和轮廓波变换对平滑分量和纹理分量进行插值;最后将插值后的平滑分量和纹理分量重新组合实现红外云图的超分辨率重建.结果表明,解决了重建图像中的平滑和伪影问题,在视觉效果和PSNR等指标上均优于其他插值算法.An infrared nephogram interpolation reconstruction scheme is proposed based on morphology component analysis. Firstly, the available low resolution infrared nephogram is decomposed into a smooth component and a texture component using morphological techniques; Secondly, the smooth component and the texture component is represented by the stationary wavelet transform and the Contourlet transform respectively. To tackle the problem of low frequency energy loss by wavelet interpolation, the effect of stationary wavelet interpolation is first improved using the method of first interpolation, followed by the smooth component and texture component being interpolated using the stationary wavelet transform and the Contourlet transform respectively. Finally, the super resolution infrared nephogram is obtained by combining the smooth component and texture component. The results demonstrate that the problem of smoothing and artifact in the super resolution reconstructed image has been solved. Both the visual effect and the PSNR index are proven better than other interpolation algorithms.

关 键 词：云图 超分辨率 形态学分量分析 平稳小波变换 CONTOURLET变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]