Abstract: To improve the robustness of watermarked images to geometric attacks, an image watermarking algorithm combining Blob-Harris feature region with Contourlet transform (CT) and singular value decomposition (SVD) was proposed. First, the Blob block region was extracted from the low-frequency sub-band coefficients of the original image after Contourlet transform and Harris corner detection was used to extract feature points. Then, the feature area was determined according to the feature scale of each feature point. A moderate non-overlapping feature area was selected and its surroundings were zero-added before normalization.Finally, the singular value decomposition was performed on the low-frequency watermark image extracted by wavelet transform, and it was repeatedly embedded in the inscribed regular quadrilateral of each normalized circular feature area.The results of simulation experiments show that in addition to a certain resistance to conventional attacks, the algorithm in this paper also has strong robustness to geometric attacks, especially translation, scaling, shearing attacks and their combined attacks, and the NC value is above 0.94.