基于SGC变换耦合边缘特征加权的图像融合算法

An image fusion algorithm based on Second Generation Curvelet transform coupled with edge feature weighting

摘要:

当前图像融合主要通过图像特征取大的方法来融合图像系数，忽略了图像边缘特征的相关性，使融合结果中存在块现象和不连续现象等不足。为此，引入二代曲波(SGC)变换，设计边缘特征加权的图像融合算法。将输入图像通过SGC变换，分解出其低频和高频系数；采用区域能量模型，计算出图像所包含的能量特征，通过均值模型，计算出图像所包含的亮度特征。利用图像的能量和亮度特征，构造双特征低频系数融合函数，获取能量特征和亮度特征都较为理想的融合低频系数。利用Sobel算子对图像的边缘特征进行检测，利用检测结果构造边缘特征加权的高频系数融合准则，获取高频系数融合结果。最后，借助二代曲波逆变换处理融合系数，获取融合图像。实验测试结果显示，相对于当前融合技术，所提算法具有更好的融合效果，其融合图像具备更好的边缘连续性以及清晰度。

Abstract:

Many image fusion methods, such as large image features to fuse image coefficients, ignore the correlation of edge features between images, resulting in block phenomenon and discontinuity in the fused image. This paper designs an image fusion algorithm with edge feature weighting based on the second generation of curved wave transform. Firstly, the input image is decomposed into low frequency and high frequency coefficients by Second Generation Curvelet(SGC) transform. Then, the region energy model is employed to calculate the energy features of the image, and the mean value model is adopted to calculate the brightness features of the image. By using the energy and brightness features of the image, the fusion function of low-frequency coefficients with two features is constructed to obtain the fusion low-frequency coefficients with excellent energy and brightness features. Sobel operator is introduced to detect the edge features of the image, and the fusion function of high frequency coefficients weighted by the edge features is constructed based on the detection results to obtain the fusion results. Finally, the fusion experiments are carried out with the algorithm. The experimental results show that the proposed algorithm has better fusion effect than current algorithms, and the edge continuity and clarity of the fusion image are better.