Pansharpening of remote sensing images using dominant pixels

Abstract

Pansharpening refers to the synthesis of super-resolution multispectral images (i.e., pansharpened images, PIs) designed to overcome the technical constraints of Earth Observation Satellites. A PI is synthesized by fusing the high-resolution chromatic information carried by the multispectral image with the high-resolution spatial information carried by the panchromatic image. This process generates precise and comprehensive data suitable for diverse applications, such as land cover mapping, urban planning, and natural resource management. Various histogram transformation techniques are utilized in pansharpening methods to improve the chromatic fidelity of PIs. However, many existing histogram transformation techniques are susceptible to chromatic distortions. In this paper, a novel pansharpening method named “Pansharpening Using Dominant Pixels” (PDP) is introduced. PDP employs a new method for histogram transformation that preserves chromatic information. Furthermore, PDP is structurally simple, easy to implement, fast, and capable of producing high-quality pansharpened images. Six Remote Sensing images and seventeen pansharpening methods were used in experiments to examine PDP’s success in generating pansharpened images. The experimental results demonstrate that PDP statistically outperforms the comparison methods, synthesizing PIs with high spectral and spatial fidelity.