改进对比度的有限离散剪切波图像融合

摘要

为了改善融合图像的对比度,保留源图像的丰富有效信息,借助有限离散剪切波(FDST)的优良特性,提出了一种基于有限离散剪切波域的图像融合算法.首先将源图像经过FDST分解,得到一系列高低频子带;然后采用改进的像素加权与区域空间频率相结合的融合方案对低频系数进行选取,高频则引入区域平均梯度对比度的融合方法;再借助FDST逆变换对融合系数进行重构获得融合图像.通过对比试验结果表明:所提算法的融合结果在主观视觉上对比度适中,图像清晰自然,且各项客观评价指标均明显改善;进一步说明了改进算法优于其他算法,大大提升了图像融合的质量.%To improve the contrast of the fused image and preserve the rich and effective information of the source image,an image fusion algorithm based on Finite Discrete Shearlet (FDST) domain is proposed,with consideration of the excellent characteristics of FDST.Firstly,the source image is decomposed by FDST,and a series of high and low frequency subband coefficients are obtained.Then,the low frequency coefficients are selected by using the improved pixel weighting together with the regional spatial frequency;and the fusion method by the regional average gradient contrast is adopted for the selection of high frequency coefficients.Then,the fused image is obtained with the fusion coefficients reconstructed by inverse transform of FDST.The contrast experiment results show that:1) The fusion result of the algorithm is clear,natural and has appropriate contrast for human subjective vision perception;and 2) The objective evaluation quality indexes are improved compared with other fusion algorithms.It is confirmed that our algorithm is superior to other fusion algorithms,which greatly enhances the quality of image fusion.