A Variational Gradient-based Fusion Method for Visible and SWIR Imagery

This paper presents a new variational gradient-based fusion method for visible and short-wave infrared (SWIR) imagery. The proposed method enables spatial enhancement and dehazing of visible imagery. Integrating gradients from SWIR imagery into visible imagery produces a single image with true color and sharp gradients. A constraint based on band correlation is included to improve the enhancement and implement dehazing. The band correlation is according to the quantitative relationship between the wavelength and the atmospheric effect caused by Rayleigh scattering. In this study, both clear and hazy Landsat ETM images are used in the experiments. By visual assessment, the gradient of the fused image is more salient than that of the original image, and the true color is well preserved. With the inclusion of the band correlation constraint, the proposed fusion method yields almost haze-free results. Quantitatively, the Metric Q of the fused images is significantly higher than that of the original images; the largest increase of the Metric Q in the experimental results is from 0.0114 to 0.0611. Moreover, for the results of the proposed method, the Metric Q increase in the visible bands declines from blue band to red band.