An adaptive video watermarking algorithm based on the DC Component is proposed in this paper. Based on the stochastically selected video frames, two dimensional 4by4 integer transform is performed on the luminance component, and then the DC coefficients are grouped to conduct one dimensional integer transform. To balance between the invisibility and robustness of watermarks, we utilize the length of the watermarks and the value of the transformed coefficients to adaptively select the groups to embed watermarks and the intensity of the coefficients modification. Watermarks are stochastically permuted and coded by LDPC to improve the robustness against attacks. Experimental results show that the proposed approach can achieve fine video quality and PSNR is higher than 50dB. Moreover, the watermarks can be extracted blindly and are robust to common video attacks. Especially, it can effectively extract watermarks under diverse compressions.