红外与可见光图像融合的多分辨率方法

Multi-resolution Scheme Appropriate to Fusing Infrared and Visible Light Images

摘要:

给出一种适合于红外与可见光图像融合的多分辨率融合方法。方法基于对比度塔形分解的图像融合方法，在融合过程中可有选择地突出被融合图像的对比度信息，达到良好的视觉效果。利用熵、交叉熵、互信息量、平均交叉熵、均方根交叉熵等参量，对多种图像融合方法的性能进行了定量评价和比较。实验结果表明，方法用于红外与可见光图像的融合十分有效。

Abstract:

Fusion of infrared and visible light image was studied.A multi-resolution image fusion scheme based on contrast pyramid(CP) decomposition was presented.In order to obtain a better fusion result,the contrast details of the images to be fused can be signalized selectively in the fusion processing.In addition,the quantitative evaluation of fusion performance for infrared and visible light image fusion was also investigated.With the parameters such as entropy,cross entropy,mutual information,average cross entropy,root mean square cross entropy,the performances of different fusion methods were evaluated and compared quantitatively.Compared with the methods based on Laplacian pyramid decomposition and discrete wavelet transform,the CP method is more appropriate to fusing infrared and visible light image.The experimental results are perfectly in keeping with the quantitative evaluation results.It shows that the evaluating parameters,evaluating standard,and the research method are reasonable and effectual.The evaluating methods presented here are also applicable to other image fusion applications.