oRGB颜色空间的夜视图像彩色融合

针对红外与可见光双波段夜视图像增强提出了基于oRGB颜色空间的彩色融合算法。选择亮度与颜色分离的对立色颜色空间,应用多尺度Top-hat变换方法将红外与可见光增强融合图像作为伪彩色图像的亮度分量,根据双波段图像的差异性并结合参考图像色调构造颜色分量;为避免线性变换的颜色传递方法容易造成灰度级压缩或过饱和情况,在分析颜色传递机理的基础上,使用非线性的直方图匹配方法实现颜色传递。实验结果表明,融合图像的颜色分布更自然,纹理细节、整体亮度和对比度等视觉特征表现更优。

Night-vision image coloration fusion in oRGB color space

A novel image coloration fusion method in oRGB color space is presented for the enhancement of thermal infrared and visible night-vision images. We chose to construct a pseudo-color image and to execute color transfer in the same oRGB color space because this new space is reasonably simple, has intuitive axes, and is well-suited for computational applications. In order to preserve detail information from the source images in the final coloration image, the luminance component of the pseudo-color image is assigned to the fused image by a multi-scale top-hat transform. The two color components of the pseudo-color image are constructed according to the differences between the dual-band images and the color statistics of the reference image. To avoid gray-scale compression or over-saturation that can be easily caused by the linear color transfer method, we applied histogram matching rather than linear transformation to the two color components in the color transfer process. Experiments show that our method produces a coloration image with a more natural color distribution. Furthermore, visual features such as salient features and overall brightness and contrast are superior to those produced by other methods.