一种改进的压缩感知图像融合算法

An Improved Compressed Sensing Image Fusion Algorithm

针对某些传统的基于压缩感知的图像融合算法在压缩感知域中进行观测值融合时,采用的融合规则可能存在的权值选取不合理不精确的问题。根据压缩感知观测值的列的完整性原理,在提出的基于压缩感知的可见光与红外图像融合算法中,对观测值的融合采用一种局部均值加权的规则。首先,将源图像进行离散小波变换,对分解得到的低频分量进行压缩感知测量,并采用局部均值加权的方式进行融合处理然后重构,对高频分量采用区域能量匹配的规则进行融合,最后将融合后的低、高频分量进行小波逆变换得到融合图像。实验采用多组可见光和红外图像,并与其他算法进行比较,结果表明该算法能在一定程度上提高融合图像的信息熵IE、平均梯度AG、标准差STD和互信息MI,取得较好的融合效果。

For some traditional image fusion algorithms based on compressed sensing,the weight selection of the fusion rules may be unreasonable and inaccurate.The fusion algorithm of visible and infrared images based on compressed sensing is proposed in this paper,which adopts a local mean weighted rule for the fusion of observations.It is based on the integrity principle of columns that compress perceived observations.Firstly,the source image is transformed by discrete wavelet transform,and the low frequency components are compressed and sensed.The method of local mean weighting is used for fusion processing and reconstruction,and the rules of regional energy matching are used for fusion of high-frequency components.Finally,the low and high frequency components are transformed by inverse wavelet transform to get the fused image.Multiple groups of visible and infrared images are used in the experiment.The experimental results show that the algorithm can improve the information entropy,the average gradient,the standard deviation and the mutual information of the fused image to a certain extent,and achieve better fusion results.