Objective: In order to balance the contradiction between invisibility and robustness of the watermarking algorithm , this paper proposed a watermarking algorithm with invisible and robust based on image-block. Method: The host image is divided into non overlapping image-blocks, the texture features and edge features of each block are analyzed by using the masking properties of human visual system and so to extract the image-blocks with good masking property , the singular value decomposition is applied to the low frequency sub-band after the wavelet transform of each selected image-block. By modifying the size relationship of the elements in the first column of the U orthogonal matrix, the watermark image with Arnold scrambling is embedded. Before the watermark extraction, geometric correction of the watermarked image with geometric distortion is realized by coordinate relations and scales features of the scale-invariant feature transform(SIFT) feature points, the approach of geometric correction restores the synchronization of the watermark. Result: Experiments on standard gray scale images, the peak signal to noise ratio values of the obtained watermarked images can reach more than 44, the normalized correlation coefficients between the original watermark image and the extracted watermark image from the watermarked image under conventional attacks and geometric attacks can mostly reach more than 0.99, which shows that this algorithm not only has good invisibility, but also has strong robustness against conventional attacks and geometric attacks. Conclusion: The image-blocks with good masking property are selected to fully ensure the invisibility of the watermarking algorithm. Especially before the watermark extraction, geometric correction is applied on the geometric distorted watermarked image to improve the ability of resisting geometric attacks , which greatly balances the contradiction between invisibility and robustness of the watermarking algorithm.