摘要：

为改善多源灰度图像的融合效果,结合有限离散剪切波变换(FDST)较高的方向敏感性和平移不变性,提出一种新的灰度图像融合算法。对经过配准后的原图像进行FDST分解,获得不同尺度和不同方向的高频子带系数与低频子带系数,对低频采用区域平均能量和平均梯度相结合的融合算法,对高频选用相对区域方差和平均梯度相结合的融合方法。利用有限离散剪切波逆变换重构得到融合图像,并对融合结果进行主观视觉和客观评价。实验结果表明,与基于小波变换的低频区域能量融合和高频区域方差的融合算法等相比,该算法能获得较好的融合效果和原图像细节描述。

关键词: 图像融合, 有限离散剪切波;平均梯度;区域方差;平移不变性

Abstract:

To improve the fusion effect of multi-source grayscale image,a new grayscale image fusion algorithm is proposed combining the shift invariance and good directional sensitivity of FDST.The original images after registration are decomposed by FDST,and the low frequency sub-band coefficients and high frequency sub-band coefficients of different scales and directions are obtained.The combination fusion algorithms of regional average energy and average gradient are used for the low frequency,and the combination fusion algorithms of relative region variance and average gradient are used for the high frequency.The fused image is reconstructed by inverse transform of finite discrete shear wave,and the result is evaluated by subjective vision and objective performance.Experimental results show that the algorithm can get better fusion results and image detail extraction compared with low frequency regional energy fusion and high frequency region variance fusion algorithms based on wavelet transform.

Key words: image fusion, finite discrete shearlet, average gradient, regional variance, translation invariance

中图分类号: 

• TP391.4