A New Adaptive Linear Interpolation Algorithm Using Pattern Weight Based on Inverse Gradient

Most of conventional interpolation methods do not consider the sufficient pattern of neighbor pixels, which cause quality degradation. Kim et al. proposed new adaptive linear (NAL) algorithm to consider patterns near the interpolated value. However, they have a critical defect which does not reflect whether each neighbor pixels influence the interpolated pixel. To remove this defect, we propose a new image interpolation method using adaptive weight based on inverse gradient. Experimental results show that the proposed algorithm exhibits a better performance than conventional algorithms in both objective and subjective criteria with a variety of images. In addition, the proposed method just needs a little computational burden compared with other algorithms.