基于NSST与IHS的红外与彩色可见光图像融合

目的 鉴于传统的红外与彩色可见光图像融合算法得到的融合图像，无法很好兼顾清晰度、对比度和色彩是否失真等，提出一种新的基于NSST和IHS颜色空间的彩色图像融合算法。方法 首先将源彩色可见光的RGB图像变换到各通道相关性最小的IHS颜色空间，分离出亮度分量和色度分量。其次对彩色可见光的亮度分量和红外图像分别进行NSST分解，对分解得到的低频系数采用基于自适应高斯模糊逻辑函数的系数选择方案，对高频系数则采用基于像素点的绝对值取大的系数选择方案，然后对经过选择的低、高频系数进行NSST逆变换，得到的融合图像作为新的亮度分量，结合已有的色度分量将其进行IHS逆变换，得到最终的RGB融合图像。结果 通过2种场景的红外与彩色可见光图像进行仿真实验，将提出的算法与LPT，SWT和NSCT等算法对比，通过主观评价和客观评价指标IE, AG, SF和SD等，可知新算法的融合结果图像场景细节最清晰，红外隐藏目标对比度最高，且色彩未出现明显失真现象，图像融合质量最高。结论 提出的算法相较于传统的红外与彩色可见光图像的融合质量，全面提升了效果，表明该算法具有优越性。

Fusion of Infrared and Color Visible Images Based on NSST and IHS

The paper aims to propose a new color image fusion algorithm based on NSST and IHS color space to solve the problem that the traditional fusion algorithm of infrared and color visible images cannot give good consideration to the fusion quality of image clarity, contrast and color distortion. Firstly, the RGB image of the source color visible image was transformed into the IHS color space of the least correlation among the channels, and the luminance and chroma components were separated. Secondly, the brightness components of color visible image and infrared image were decomposed by NSST. The low-frequency coefficients obtained by NSST decomposition were selected based on adaptive Gaussian fuzzy logic function, and the high-frequency coefficients were selected based on the absolute value of the pixel points. Then the selected low and high frequency coefficients were inversely transformed by NSST, and the fused image was taken as a new luminance component, and then inversely transformed back to RGB space by combining the existing chroma components to get the final fused image. The fusion of infrared and color visible images in two scenes was simulated. Compared with LPT, SWT and NSCT, subjective evaluation and objective evaluation indexes such as IE, AG, SF and SD showed that the scene details were the clearest, the infrared hidden objects had the highest contrast, and the color was not distorted obviously. The image fusion quality was the highest. Compared with the traditional infrared and color visible image fusion, the proposed algorithm integrally improves the quality of the fused image, showing the superiority of the algorithm.