The work aims to propose a color image blind watermarking algorithm resistant to rotation attacks in view of such problems as the complex color image blind watermarking algorithm with poor anti-rotation ability, and combined with Schur decomposition and Radon transform. Firstly, the proposed algorithm extracted the U channel image of color host image, and then divided the low frequency components of U channel image after NSCT transform into blocks to decompose each pixel block into symmetric and anti-symmetric matrix. Finally, the Schur decomposition of each symmetric matrix was conducted, and the binary watermark was embedded in the maximum eigenvalue of the decomposed upper triangular matrix. For the extraction of watermark information, Radon transform was used to extract the watermark after the geometric correction of watermark image. The proposed algorithm extracted watermarks with high similarity and its NC value reached 0.9414. The structural similarity was high and its SSIM value reached 0.9183. The watermarking algorithm is highly robust against conventional signal processing attacks and anti-rotation attacks.