Currently, we are in the area of the digital big data world. Every day, people share billions of images over Internet from various sources. Ensuring images’ authorship rights and illegal modification and/or reproductions become a challenge. Their protection must meet current and secure development needs of multimedia applications. In this paper, we propose an efficient watermarking scheme based on combining Imperialistic Competitive Algorithm (ICA), Speed Up Robust Features (SURF) points and Discrete Cosine Transform (DCT). The high complexity and inappropriate embedding area could lead to weak performance of watermarking schemas. To address these problems, an ideal watermarking embedding area was proposed using ICA to provide an optimized cost and complexity. The watermark was embedded into the DCT coefficients of the selected embedding areas. The watermark was extracted in a blind way to ensure color image’ authentication. Experimental findings demonstrate that the proposed approach achieves promising performance in terms of robustness and imperceptibility in almost cases. The average PSNR exceeds 61db regarding the imperceptibility. For the robustness measured by the correlation coefficients between the original watermark and the extracted is close to 1 in almost cases.