图像融合的非负线性混合模型与算法研究

本文提出了一种适用于像素级图像融合的非负线性混合模型,该模型的意义在于任何像素级图像融合问题都可以用此形式化模型来表示.针对实际融合问题,根据实际情况对模型进行具体化或者简化,就能够得到具体融合问题的融合算法.在此基础上,根据图像融合的实际问题,本文给出了该模型的一种简化模型,并对简化模型的求解问题进行了深入研究.通过将简化后的模型转化为具体的优化问题,给出了一种快捷的图像融合迭代算法.所求出的图像融合结果综合了源图像中的特征信息,使得融合图像的特征信息损失较少,更接近于理想的真实图像.该方法的收敛性可以保证,是对现有图像融合方法的补充,能够应用于多种实际的图像融合处理中.大量的实验并与其他的融合算法的比较分析表明了本文算法的有效性和优越性.

Non-negative Linear Combination Model and its Algorithm in Image Fusion

This paper presents a new non-negative linear combination model applicable to image fusion at the pixel lev- el, which can be used to represent the pixel-level image fusion. The algorithm for a fusion can be achieved by reifying or simplifying the non-negative linear combination model. In this method, a simplified model has been given and studied in detail to find its solution in this paper. And a convenient iteration algorithm for this simplified model is obtained by converting the problem into an optimization problem. The resultant image integrates the feature information in the origi- nal images, and reduces the loss of feature information in fused image so that it approximates nearer to the true im- age. Simple in implementation and guaranteed in convergence, the algorithm is bound to be a useful supplementation in image fusion algorithms available and is applicable to various applications in image fusion processing. Lots of experi- ments and their comparisons with other fusion algorithms show the new method's effectiveness and superiority.