基于边缘特征和方向相关性的图像融合方法

An Image Fusion Method Based on Edge Characteristics and Directional Correlation

不同传感器获得的多幅图像对同一场景的描述具有信息的冗余性和互补性,在小波图像融合的基础上提出了一种基于边缘和方向相关性的图像融合算法。对参加融合的两幅图像进行小波多尺度分解,对高低频子带系数分别采取不同的融合方法;根据高频分量不同的小波方向特性,采用基于方向相关系数的融合规则,而对低频分量采用能量和边缘特征相结合的融合方法;通过小波重构得到融合图像。实验结果表明,融合图像的视觉效果较好,客观评价指标均有提高。

A new image fusion rule based on edge characteristics and directional correlation was proposed to fuse the multiple images that there is some redundancy and complementarity between different images of the same scenarios .The wavelet multi-scale decomposition technique is applied to decompose these images that would be fused first .Then an algorithm was used to calculate the fusing coefficient of high frequency subband that is different from the algorithm for low frequency subband .The directional corelation was proposed to calculate the fusing coefficient for every coefficient of the high-frequency in different directions,and the low-frequency approximation component was determined by image edge characteristics and energy .Final fusion image is obtained by wavelet reconstruction .The good visual effect of the simulation experimental results is obtained .All various indicators of objective evaluation index is increased .This method is proved to be effective .