Contrast enhancement is essentially needed in mapping preprocessing due to low average brightness and low contrast of
original remote sensing images. The frequently used methods improve the visual quality of remote sensing images while
compress gray ranges of skin-tone. As a result, shape and texture of these oversaturated regions are missing or distorted.
The proposed method named GL-Enhancing not only enhance the contrast of remote sensing image, but also keep shape
and texture. It is found that GL-Enhancing performs better than Equalization, Linear-Stretch, Gauss and SquareRoot
methods in QuickBird 321bands true color image experiment. GL-Enhancing is efficient in mapping preprocessing. The
proposed algorithm is easy to implement.
Among most of current Pan-sharpening methods, resampling is generally required to make panchromatic (Pan) and
multispectral (MS) images matched correctly pixel by pixel. However, few methods have focused on spectral distortions
caused by shape distortions of real features during resampling. This paper proposes a new Pan-sharpening algorithm
based on the gray and spectral relationships between Pan, MS and the fused images. In the algorithm, Pan-sharpening is
defined as an optimization of a linear overdetermined system. It takes Pan and original MS images as input datasets
without resampling. The Least square technique is applied to calculate the optimum values (quality fused images).
QuickBird image datasets are tested, and the results are compared with the fused images of IHS, PCA and Gram-Schmidt
using interpolated MS image. The result shows that the proposed method is more efficient than IHS, PCA and
Gram-Schmidt in preserving spectral characteristics and increasing spatial resolution, especially for high spatial
resolution ratio (SRR > 4:1, spatial resolution ratio is the ratio of the spatial resolution of MS image to that of Pan image.) images.