By using intensity probability density and color invariance model, a novel feature-based color image watermarking method is proposed in this paper. Firstly, the uniform and robust feature points are extracted in the color invariance model by utilizing modified speeded up robust features (SURF) detector, in which the intensity probability density gradient is used instead of intensity gradient. Then, the affine invariant colored local ellipse regions (LERs) are constructed adaptively according to the variation of local intensity probability density. Finally, the digital watermark is embedded into the affine invariant colored LERs based on statistical theory and image histogram invariant theory. Experimental results support the potential of the proposed approach.