有限离散剪切波域的红外可见光图像融合

发布时间：2019-05-21 19:07

【摘要】：针对目前图像融合过程中的不足之处,结合有限离散剪切波具有高的方向敏感性和抛物尺度化特性,提出了一种有限离散剪切波变换下的图像融合算法。首先对严格配准的多传感器图像进行有限离散剪切波变换,得到低频子带系数和不同尺度不同方向的高频子带系数;然后对低频子带系数采用全局特征值和像素点之间的差异性与区域空间频率匹配度相结合的融合算法,高频方向子带系数采用方向权重对比度与相对区域平均梯度和相对区域方差相结合的方案;最后通过有限离散剪切波逆变换得到融合图像。实验结果表明,与其他的融合算法相比较,本文算法不但有良好的主观视觉效果,而且3幅图像的客观评价指标分别平均提高了0.9%、3.8%、3.1%,2.6%、3.8%、2.9%和1.5%、125%、59%,充分说明了本文融合算法的优越性。
[Abstract]:In view of the shortcomings of the current image fusion process, an image fusion algorithm based on finite discrete shear wave transform is proposed, which combines the high direction sensitivity of finite discrete shear wave and the characteristics of parabola scaling. Firstly, the finite discrete shear wave transform is carried out on the strictly registered multi-sensor images, and the low frequency subband coefficients and the high frequency subband coefficients in different scales and different directions are obtained. Then the low frequency subband coefficient is combined with the difference between global eigenvalues and pixels and the regional spatial frequency matching degree. The high frequency direction subband coefficient adopts the scheme of combining the contrast of direction weight with the average gradient of relative region and the variance of relative region. Finally, the fusion image is obtained by finite discrete shear wave inverse transform. The experimental results show that, compared with other fusion algorithms, the proposed algorithm not only has a good subjective visual effect, but also improves the objective evaluation index of the three images by 0.9%, 3.8%, 3.1% and 2.6%, respectively. 3.8%, 2.9% and 1.5%, 125% and 59%, which fully demonstrates the advantages of the fusion algorithm in this paper.
【作者单位】：西安建筑科技大学理学院;
【基金】：陕西省自然科学基金资助项目(No.2015JM1024;No.2013JK0568)~~
【分类号】：TP391.41

【参考文献】