Pattern　matching　has　found　wide　application　in　signal　processing,　computer　vision,　image　and　video　processing.　Recently,　the　Incremental　Dissimilarity　Approximations　(IDA)　algorithm　is　successfully　applied　for　pattern　matching.　This　paper　proposes　a　method,　called　Black　Sheep　algorithm,　which　utilizes　unmatched　sub-block　to　eliminate　all　blocks　that　contain　this　sub-block.　Then　Black　Sheep　rejection　condition　and　Black　Sheep　threshold　are　proposed　to　further　relieve　the　burden　of　computation.　Experiment　results　show　the　high　efficiency　of　the　proposed　algorithm.　©　2010　IEEE.