Expanded Q4 quality assessment for pan-sharpened multispectral image

Abstract

ABSTRACT In this paper, we took into account both the spectral info rmation and the spatial information and estimated how well the needed information contained within the multispectral (MS) and panchromatic (PAN) images was represented by the pan-sharpened image. Based on that, we proposed a new qua lity index which could be seen as an expanded index of the global quality measurement Q4. In our method, we first m easured the spectral information preserving quality between the MS image and the fusion result. Then, we constructed a virtual spatial detail image considering the spatial resolution ratio between the source MS image and the PAN image, and also extracted the detail image contained in the merged image using the same technology, followed by a spatial information preserving quality index calculated from these two detail images. At last, we integrated the two indices by means of weighted addition determined by fusion model. To illustrate the superiority of this new index, we took experiments on two pairs of ZY-2 PAN and ASTER MS (1 2 3 bands) remote sensing imageries, and adopted the tradeoff FIHS fusion method in which the tradeoff parameter was set to different values standing for different fusion models. After using the propos ed index to assess the quality of fusion, we think that the new index is compliant with subjective evaluati ons and could therefore be used to compare different image fusion or to find the best parameters for a given fusion mode l. Finally, we gave an experiential weight parameter of the quality index while assessing the tradeoff FIHS fusion with im ages from these two sensors by the author’s experiments. Keywords: pan-sharpen, non-reference quality assessment, fusi on model, quaternion algebra, weight parameter