基于NSCT变换和压缩感知的图像融合

西安科技大学 通信与信息工程学院,陕西 西安 710054

图像融合; 压缩感知; 非下采样contourlet变换; 可见光; 红外

Image fusion based on NSCT trasformation and compressive sensing
WU Yan-hai,ZHANG Ye,MA Meng-xin

(College of Communication and Information Engineering,Xi'an University of Science and Technology,Xi'an 710054,China)

image fusion; compressive sensing; Non-subsample contourlet transform; infrared images; visible light images

DOI: 10.13800/j.cnki.xakjdxxb.2015.0413

备注

针对NSCT变换分解得到的各个方向子带稀疏度不同的问题,文中提出了1种基于改进的CS_NSCT图像融合方法。首先对待融合图像进行NSCT分解,接着对得到的高频分量采用自适应的压缩感知方法进行压缩,并在压缩域融合后重构; 对低频分量采用DCT能量准则融合,最后对融合后的高低频分量进行NSCT重构。仿真实验结果表明,文中方法在减少了数据量的同时有效提高了图像的熵值、标准偏差、平均梯度等指标。

Through NSCT transformation,an image will be decomposed into a low-pass sub-band and K-direction sub-bands,but the sparsity is different for each direction.Therefore,this paper proposes an improved CS_NSCT way for fusion of infrared and visible light images.First,making a NSCT decomposition to the infrared image and visible light image,next doing compression to high frequency sub-bands by adaptive compressive sensing,and after that to fuse them.For the low-pass sub-band,it uses block DCT of high frequency energy rule to fuse them.Finally,it gets fusion image by reconstruction of compressive sensing and inverse NSCT transformation for data which has been fused.The simulation shows that it not only improves parameters,such as entropy,and standard deviation,average gradient,but reduces the amount of data effectively.