We　proposed　an　image　signature　method　for　content　authentication,　which　applies　a　hierarchical　approach　to　construct　an　image　signature.　In　the　first　level,　DWT　and　DCT　are　used　to　extract　image　features;　then　these　features　are　encrypted　by　using　sub-keys　that　are　generated　by　a　cryptographically　hash　function.　In　the　second　level,　Karhunen-Loeve　transformation　is　used　to　reduce　the　signature　length.　The　main　features　of　the　proposed　method　are　as　follows:　(i)　It　achieves　a　trade-off　between　robustness　and　tampering　sensitivity.　(ii)　It　provides　a　tool　for　image　tampering　detection　and　tampering　localization.　(iii)　It　can　be　used　to　detect　the　thumbnail　of　the　large　image　to　improve　detection　efficiency.　(iv)　It　provides　the　compact　signature,　and　the　signature　length　is　independent　of　the　image　size.　Experimental　results　show　that　proposed　method　is　robust　for　content-preserving　manipulations　such　as　JPEG　compression,　adding　noise,　filtering　and　Gamma　correction,　etc.