Abstract: This study proposes a new color image super-resolution method in an adaptive and robust framework. In the framework, three channels of color images are reconstructed, respectively. Total variation regularization is used for edge preservation of single-channel component. In order to overcome the shortcoming of the total variation model which often leads to generate an undesirable staircase effect, a spatial adaptive function was designed based on locally structural features of images. This function was used to couple the low-order total-variation model with the beyond digital total-variation model. Then, an adaptive variational color super-resolution algorithm of image sequences was proposed to preserve image edge structures and remove the color staircase effect. Experimental results on real images were presented which demonstrate the effectiveness of the proposed method.