An Improved DCT interpolation using Bilateral filter
Jagyanseni Panda, Sukadev Meher · 2021 IEEE 18th India Council International Conference (INDICON) · 2021
The image interpolation method is used to improve the resolution of low resolution (LR) images. Several methods, such as polynomial and edge-directed methods, are proposed, but interpolated images are associated with blurring and ringing artifacts. As a result, one pre-processing method based on a bilateral filter is proposed. The bilateral filter is applied to the LR image, which blurs it using spatial and range filtering. The mask is determined by subtracting the LR image from the filtered image. The sharpened LR image is obtained by combining the LR image with a weighted mask image using the unsharp masking concept. Finally, using the discrete cosine transform (DCT), the restored high resolution (HR) image is obtained by interpolating the sharpened LR image. The high frequency as well as texture details are preserved as a result of this pre-processing. In terms of PSNR and SSIM, the proposed method outperforms both subjective and objective measures.