The color filter array (CFA) is a cost-effective structure widely adopted in modern digital still cameras (DSC). Because some color information has been filtered out, interpolation process is required to retrieve the original image. In literature, bilinear interpolation is commonly used due to its simplicity. No cross-channel computations are necessary in bilinear interpolation. By taking into consideration of inter-channel correlation, Pei and Lam improved the bilinear interpolation without increasing much of computational complexity. Both approaches however suffer certain edge-aliasing effect by reason of their average (or lowpass) nature. In order to eliminate such edge-aliasing, a new interpolation method is proposed in this work. By excluding extreme neighbor pixels before the average operation, the edge can now be better preserved. Experiment results show that the proposed interpolation method not only preserves the object edges visually but generally gives smaller mean-square errors.