A　spectral　imagery　compression　algorithm　based　on　spectral　prediction　and　wavelet　transform　is　presented.　Utilizing　spectral　prediction　and　2D　wavelet　transform　respectively,　the　special　and　spectral　redundancy　is　removed.　Then,　combined　with　band　reordering　and　quadtree-based　subband　by　subband　wavelet　image　coding　method,　the　experimental　results　of　the　proposed　scheme　show　obvious　advantages.　At　the　same　compression　ratio,　the　average　PSNR　of　reconstructed　images　is　3.69　dB　higher　than　that　of　JPEG2000.　Meanwhile,　with　the　anti-error　and　multi-resolution　decoding　features,　it　is　suitable　for　the　application　in　remote　sensing　platforms.