The near-infrared band of the electromagnetic spectrum has become an important tool for enhancing image quality. Commonly, outdoor color images are degraded by bad weather conditions that lead to a loss of contrast and fine details in color images since light scattering produces attenuation and smoothing effects. Despite the fact that current Visible-NIR fusion methods achieve image enhancement features, some issues like edge preservation and color oversaturation still need to be addressed. In this work, a method that performs a selective Visible-NIR fusion of the most relevant image structures through top-hat transform is proposed. The performance of the method is evaluated by quantifying the new information added to the image and the change in color. Experimental results show a high degree of detail in preserving the edges while maintaining the color of the image. Moreover, the proposed method demonstrates that the image quality improvements were not significantly affected by a change in the color space.

Automated classification systems have allowed for the rapid development of digital large sky surveys. Such systems increase the independence of human intervention in the analysis stage of star and galaxy classification. Artificial neural networks, hierarchical classifiers and ensembles of classifiers have been used as the methods of classification in these systems. This paper investigates the development of an automated classification system for galaxies in astronomical images based on the method of sparse representation. The dependency of classification based on image enhancement by the alpha-rooting, heap-, and paired-transforms is secondarily investigated.