An Edge-Guided Weighted Image Interpolation Algorithm

For the problem of poor performance of linear image interpolation and high complexity of nonlinear image interpolation, an edge-guided weighted image interpolation algorithm is proposed. Firstly, the edge feature of the low-resolution image is detected by the Sobel operator, and the edge strength and direction of the high-resolution image are calculated by the bilinear interpolation algorithm. Then, the warped distance of the reference pixels and weighted Lanczos kernel are calculated by the edge features of the high-resolution image. Finally, the reference pixels are weighted to obtain the resulting high-resolution image. The simulation results show that the proposed interpolation algorithm is more accurate than traditional interpolation algorithms, and the computing time of the upscaling by a factor of 2 for the
$\mathbf{256}\times \mathbf{256}$
image with the proposed algorithm is less than 50ms, which is attractive for real-time communication and image applications.