有效的红外与可见光图像融合方法研究

Research on effective fusion method for infrared and visible images

针对红外图像可视化程度弱、对比度低的问题,提出一种基于轮廓小波变换和区域能量的红外与可见光图像融合算法。首先进行多尺度小波分解,然后进行多方向滤波;引入循环平移方法来消除伪吉布斯失真;采用基于区域的能量融合规则,重构变换系数得到最终融合结果;最后用信息熵、信噪比等指标来评价融合的性能。实验表明,该方法不论在客观评价还是在主观评价指标上都优于其他融合方法,提高了融合图像的视觉效果,可以得到更加清晰的融合图像。

For visualization and contrast of IR images are weak, this paper proposed an algorithm for infrared and visible images fusion using wavelet-based contourlet transform （WBCT） and region energy. Used the WBCT to perform a multi-scale decomposition of each image in the first step, which was used to perform a multi-direction filter in＇the second step. Introduced cycle spinning to remove the effect of pseudo-Gibbs phenomena. Replaced the fusion on the whole images domain by using region-energy-based approach. To get the final object WBCT coefficients of fused image were re-constructed using multiple operators according to different fusion rules. Evaluated the performance of image fusion method using some criteria including entropy and signal-to-noise ratio and so on. The experimental results show that it is effective in proving quality of fusion image , which not only get higher object assessment index score, but also get a more clearly fusion image in perceptual than other methods.