An efficient fusion technique for quality enhancement of remotely sensed images

摘要

Remote-sensing satellites provide both high-resolution panchromatic and low-resolution multi-spectral images. In this paper, a pixel-level multi-sensor image fusion technique is proposed for improving the spectral quality of the fused multi-spectral and panchromatic remote-sensing images. The proposed fusion technique integrates both the intensity, hue, and saturation (IHS) and the discrete wavelet frame transform (DWFT) techniques. IHS fusion technique can provide high spatial quality and the DWFT fusion technique is both aliasing free and translation invariant, but the color distortion in both techniques is noticeable. The proposed fusion technique can improve the spectral discrepancy of the fused images, while preserving the spatial quality to an acceptable level. Landsat-5 (TM) with Spot (Pan), Landsat-7 (ETM+), and IKONOS panchromatic and multi-spectral images have been fused using the proposed technique. The statistical analysis shows that this technique improves the fusion quality compared to the other known fusion techniques such as the conventional IHS, discrete wavelet transform (DWT), DWFT, and the integrated IHS and DWT.