基于Contourlet变换和全变差的图像融合

Image Fusion Method Based on Contourlet Transform and Total Variation

针对基于小波和全变差的图像融合方法的缺点和不足,提出了一种基于Contourlet变换和全变差算子的含噪图像融合方法,这种方法在融合图像的同时,也对图像进行了去噪处理.首先对图像进行Contourlet变换,对得到的低频子带用分块主成分分析的方法进行处理,以估计传感器的退化算子,然后进行低频部分的融合;将高频系数归一化的全变差算子的模作为权因子进行高频部分的融合,以经Contourlet逆变换后得到的融合图像作为初始值,经过迭代演化,最终得到了去噪的融合图像.实验结果表明,该方法得到的图像视觉效果更好,信息熵更高.

To conquer the weakness of existing image fusion method based on the wavelet transform and totalvariation（TV）,a novel algorithm for fusion images with noise by taking advantages of contourlet transform and TVis proposed. This method fuses the multiple sensor images and deal with the noise at the same time. The contourlettransform is firstly used to decompose the multi-sensor source images into a low frequency subband and a series ofhigh-frequency subbands. Then,fusion rules are applied to low and high frequency subbands respectively. Theblock-principle component analysis（B-PCA）is introduced to estimate sensor gains in the low-frequency subband.Thus,a TV based the approach is then used iteratively to obtain the fused image. Numerical simulations are carriedout to validate the effectiveness of the proposed method.