Abstract: According to the contradiction between the robustness and imperceptibility of database watermarks，a robust blind watermarking algorithm based on wavelet transformation is put forward.By screening programming，the algorithm screens the candidate attributes which can embed watermarks，and splits and re-ranges the subset，and then respectively DWT transforms the subset data and the scrambling watermarking images.With the condensation algorithm，the low-frequency part of the watermark is condensed and embedded into the high-frequency part of the subset data，so as to achieve the mergence of watermarks and original data.Theoretical analyses and experiments show that this algorithm possesses strong robustness and good invisibility.