Search

This paper presents an image fusion method suitable for pan-sharpening of multispectral (MS) bands, based on nonseparable multiresolution analysis (MRA). The low-resolution MS bands are resampled to the fine scale of the panchromatic (Pan) image and sharpened by injecting highpass directional details extracted from the high-resolution Pan image by means of the curvelet transform (CT). CT is a nonseparable MRA, whose basis functions are directional edges with progressively increasing resolution. The advantage of CT with respect to conventional separable MRA, either decimated or not, is twofold. Firstly, directional detail coefficients matching image edges may be preliminarily soft-thresholded to achieve a noise reduction that is better than that obtained in the separable wavelet domain. Secondly, modeling of the relationships between high-resolution detail coefficients of the MS bands and of the Pan image is more fitting, being accomplished in the directional multiresolution domain. Experiments are carried out on very-high-resolution MS + Pan images acquired by the QuickBird and Ikonos satellite systems. Fusion simulations on spatially degraded data, whose original MS bands are available for reference, show that the proposed curvelet-based fusion method performs slightly better than the state-of-the art. Fusion tests at the full scale reveal that an accurate and reliable Pan-sharpening, little affected by local inaccuracies even in the presence of complex and detailed urban landscapes, is achieved by the proposed method.

This paper addresses the modeling of wavelet coefficients for multispectral (MS) band sharpening based on undecimated multiresolution analysis (MRA). The coarse MS bands are sharpened by injecting highpass details taken from a high-resolution panchromatic (Pan) image. Besides the MRA, crucial point is modeling the relationships between detail coefficients of a generic MS band and the Pan image at the same resolution. Once calculated at the coarser resolution, where both types of data are available, such a model shall be extrapolated to the finer resolution in order to weight the Pan details to be injected. The goal is that the merged MS images are most similar to what the MS sensor would collect if it had the same resolution as the broadband Pan imager. Three injection models embedded in an ''a trous'' wavelet decomposition will be described and compared on a test set of very high-resolution QuickBird MS+Pan data. Two models work on approximation and detail coefficients, respectively, and provide a certain degree of unmixing of coarse MS pixels. The third model is based on spectral fidelity of the merged image to the (resampled) original MS data, that is, no unmixing is attempted. It is much simpler than the other two because it does not require calculation of local statistics. Fusion comparisons on spatially degraded data, of which higher-resolution true MS data are available for reference, show that the former two models yield better results than the latter, in terms of both radiometric and spectral fidelity. The latter, however, yields a reliable sharpening unaffected by local artifacts, regardless of landscape complexity. When local statistics of wavelet coefficients are used, the model estimated on approximation yields slightly better but considerably stabler results than that calculated starting from bandpass details.

This paper addresses the modeling of wavelet coefficients for multispectral (MS) band sharpening based on undecimated multiresolution analysis (MRA). The coarse MS bands are sharpened by injecting highpass details taken from a high resolution panchromatic (Pan) image. Besides the MRA, crucial point is modeling the relationships between detail coefficients of a generic MS band and the Pan image at the same resolution. The goal is that the merged MS images are most similar to what the MS sensor would collect if it had the same resolution as the broadband Pan imager. Three injection models embedded in an "a trous" wavelet decomposition will be described and compared on a test set of very high resolution QuickBird MS+Pan data. Two models work on approximation and detail coefficients, respectively, and provide a certain degree of unmixing of coarse MS pixels. The third model is based on spectral fidelity of the merged image to the (resampled) original MS data, that is, no unmixing is attempted. Fusion comparisons on spatially degraded data, of which higher-resolution true MS data are available for reference, show that the former two models yield better results than the latter, in terms of both radiometric and spectral fidelity. The latter, however, yields a reliable sharpening unaffected by local artifacts, regardless of landscape complexity. When local statistics of wavelet coefficients are utilized, the model estimated on approximation yields slightly better but considerably stabler results than that calculated starting from bandpass details.