针对当前图像融合算法多采用一个源图像的子带能量函数作为融合系数, 导致融合图像质量不理想, 提出一种基于离散Curvelet变换与自适应能量模型的多聚焦图像融合算法。利用离散Curvelet对源图像进行多尺度分解, 以获取图像的低频子带和高频子带; 将低频子带分割成子块, 利用离散Curvelet系数来构造平均能量函数, 以此建立自适应能量模型; 引入信息熵模型, 对高频子带所包含的信息量进行度量; 通过高频子带所含信息量特征和清晰度特征, 完成图像高频子带的融合。实验结果表明: 与当前多聚焦图像融合算法相比, 所提算法具有更高的融合质量, 其输出图像具备更好的细节表现能力。

The current widely used sub-band energy function of image fusion algorithm, usually only one sub-band energy function as a fusion coefficient, has led to lack of ability and robust performance.A multi focus image fusion algorithm based on discrete Curvelet transform coupled with adaptive energy model is proposed. The discrete Curvelet is used to decompose the source image quickly and finely, so as to obtain the low frequency sub-band and the high frequency sub-band of the image. The low frequency subband is divided into sub blocks, discrete Curvelet coefficients to construct the average energy function of the pixels in each segment corresponding to the establishment of adaptive energy model by averaging the energy function. An information entropy model is introduced to measure the information contained in the high frequency subbands. The spatial frequency model is used to measure the resolution of the high frequency subbands. By the information characteristics and the definition features of the high frequency sub-band, the high frequency subbands are fused. Simulation results show that compared with the current multi focus image fusion algorithm, the proposed algorithm has good performance in detail, and the fused image has better visual effects.