Approximate　description　of　the　observation　model　has　severely　prevented　performance　improvement　of　the　existing　Super　Resolution(SR)　algorithms.　To　solve　this　problem,　a　new　weight-matrix　based　blind　super　resolution　algorithm　is　proposed　in　this　paper.　A　new　observation　model　based　on　a　motion　compensate　matrix　and　a　weight-matrix　is　defined　first.　Then　it　is　introduced　into　the　traditional　framework　of　Maximum　a　Posterior　(MAP),　where　the　problem　of　super　resolution　is　converted　to　joint　estimation　of　the　weight　matrix　and　the　High　Resolution　(HR)　image.　Evaluated　by　both　still　and　active　image　sequences,　the　proposed　algorithm　can　model　the　degrading　process　of　the　acquired　low　resolution　images　accurately.　It　shows　obvious　performance　improvement　compared　with　the　traditional　MAP　based　super　resolution　algorithm.　For　several　cases,　it　has　even　exceeded　the　results　when　the　observation　model　is　known.