New adaptive interpolation scheme for image upscaling

Preserving edge structures and image details simultaneously is considered the main challenge for image interpolation techniques that produce high-resolution images from their low-resolution counterparts. Two variants of a new adaptive interpolation scheme are proposed in this paper. In the proposed scheme for better interpolation of natural images, a new estimation mechanism that utilizes discontinuities in blocks around missing pixels is devised to discriminate strong edges. Strong edge pixels are obtained by using amended error linear interpolation and cubic convolution interpolation. Adaptive interpolation weights determined by inverse intensity distances in local windows are used to produce non-strong edge pixels based on local image structure. The proposed amended linear interpolation and cubic convolution interpolation exhibited approximately comparable performances. Simulation results on different types of images, including natural, texture, and cartoon images, demonstrate that, compared with other state-of-the-art algorithms, the proposed algorithm can generate better visual quality of the magnified images with higher peak signal-to-noise ratio (PSNR), structural similarity (SSIM), feature similarity (FSIM) index, and reasonable execution time.