In　this　paper,　we　describe　an　algorithm　to　decompose　multi-lead　ECG　data　simultaneously　into　different　linear　expansions　of　the　same　atoms　that　are　selected　from　a　redundant　dictionary.　These　waveforms　are　chosen　in　order　to　best　match　the　signal　structures.　The　proposed　approach　is　a　modification　to　the　Simultaneous　Orthogonal　Matching　Pursuit　(SMP),　and　it　is　a　general　procedure　to　compute　adaptive　multiple　signals　representations　with　a　dictionary,　a　combination　of　Hermite　functions　and　Gabor　functions.　By　comparison　to　SMP,　much　less　atoms　are　selected　to　present　signals　in　our　algorithm,　since　two　adjacent　atoms　will　be　replaced　by　their　optimal　compounding　if　some　condition　is　satisfied.　Therefore,　the　compression　ratio　of　Multi-ECG　can　achieve　better　performance　by　our　method　than　SMP.　©　2010　IEEE.