Lossy Compression of Multispectral Remote-Sensing Images Through Multiresolution Data Fusion Techniques

This work reports about a specific application concerning lossy compression of multispectral (MS) and panchromatic (P) images collected by spacebome platforms. Generally, the former is a set of narrow-band spectral images, while the latter is a single broad-band observation imaged in the visible and near-infrared wavelengths. Since high resolution spectral observations having high SNR are difficult to obtain, and especially to transmit, the P image, having resolution typically four times that of MS, but slightly lower SNR, is added to the MS data and used with the main purpose of expediting both visual and automatic identification tasks, possibly through an integration (merge) with the lower resolution MS data. Whenever MS data at the same resolution of the P data and with adequate SNR were hypothetically available on board, the bottleneck of downlink to receiving stations would impose severe limitations in the bit rate, so that a lossy compression would be mandatory. The consequence of the loss of information is a distortion, both radiometric and especially spectral, which may be easily quantified. Experimental results were carried out on simulated SPOT 5 data, constituted by three 10m MS bands (XS) and one 2.5m P band, all S b/pel. of a highly detailed urban area. The main result is that, for cumulative bit rates larger than 4.2 b/pel (at a 2.5m scale), lossless compression of the three 10m XS bands and near-lossless compression of the 2.5m P band, followed by XS + P fusion of the data decoded at the receiving station, is preferable, in terms of radiometric and especially of spectral distortion, to lossy compression of the three 2.5m XS bands, even if data with such a resolution were hypotetically available on board.