基于目标提取与NSCT的图像融合技术研究

针对红外图像目 标突出、背景对比度差、可见光图像分辨率高以及对热目标不敏感且隐蔽目标无法发现等问 题,本文 提出了一种能够突出两种图像的优势信息且不降低图像分辨率的图像融合方法。首先采用 Renyi熵提取红外目标,并通过非下采样Contourlet变换(NSCT)对可见光图像进行自适应增 强；然 后将图像增强后的可见光图像与目标图像进行融合,得到融合图像。通过对比,本文融合图 像的视觉 效果明显优于其他方法的融合图像。最后,采用多种指标进行评价。结果表明,基于目标提 取的图 像融合方法可以有效突出目标信息,提高图像的亮度和对比度,特别适于低能见度下的目标 检测与图像增强。

Study of image fusion technology based on object extraction and NSCT

Redundant information can be integrate d effectively and visual effect is enhanced through the multi-sensor image fusion techniques.Due to the prominent target and lower back ground contrast for infrared images,while high resolution but insensitive to hot and concealed targets for vi sible images,an effective image fusion method,which can highlight the two images without reduc ing the image resolution,is presented in this paper.Firstly,targets are extracte d from the infrared image by Renyi entropy,and the visible image is enhanced adaptively through non-subsample Contourlet transform (NSCT). Then the enhanced visible image is fused with the target infrared image.Finally,two different sets of infrared and visible images are used for image fusion test by different meth ods.By contrst,the fused image has significantly improve brightness and higher c ontrast than other images,and has better visual effect.In order to objectively e valuate the proposed method,many evaluations,such as information entropy (IE),st andard deviation (STD),spatial frequency (SF),mutual information (MI) and runnin g time,are adopted to make comparision with other fusion methods.The results sho w that the IE,STD,SF and MI values of the proposed method. obtained are significantly higher than those of the other methods.The proposed method has longer running time than the Laplace,wavel et transform and Renyi entropy methods,but shorter running time than other metho ds.So the proposed method is an effective image fusion one.