针对现有融合方法不易根据后续处理目的对融合规则进行自适应调整，不同方法的优点不易综合的问题，提出一个基于数据同化和差分进化算法的图像融合框架.在该框架下，将基于非采样下的Contourlet变换作为模型算子，离散小波变换作为观测算子，可根据后续处理对图像各个属性指标值的依赖程度确定各个属性指标的权重，构造由图像各个属性评价指标的加权和所组成的目标函数，再利用差分进化算法来优化目标函数，从而获取更合适的图像.二组实验从视觉效果和量化指标(标准方差、平均梯度、熵、空间频率及均方根交叉熵)两方面验证了该框架的有效性.

In order to solve the problem that fusion rules of existing image fusion methods can not be adjusted adaptively according to successive application target of the fusion image, and advantages of different fusion algorithm can not be integrated, an image fusion framework based on data assimilation and differential evolution algorithm (DE) is proposed. In this framework, nonsubsampled contourlet transform is used as model operator and discrete wavelet transform with DBSS(2,2)as observer operator. The objective function is composed of weight sum of indices, which are determined according to their relations with following application, and DE is employed to obtain proper image. Two groups of experiments with the help of the quantitative parameters(entropy, average gradient, standard deviation spatial frequency and interactive entropy of rms)and visual analysis verify feasibility of the framework.