An Efficient Image Interpolation Using Edge-error Based Sharpening

Image up-scaling is a technique for converting low resolution (LR) images to high resolution (HR) images of different size. While performing interpolation, there is some blur occurring in the high frequency (HF) region of image. So to overcome the above problem, a novel pre-processing technique based on the concept of unsharp masking is proposed. Here, the LR image is up-scaled using Lanczos interpolation method. Using the laplacian filter, the edge of the LR image and up-scaled images are calculated correspondingly. Then, the blurred HR edge is downsample and subtracted from LR edge, to produce a sharpened edge-error image. The LR image is added to sharpened edge-error image using the concept of unsharp masking. Again, this sharpened LR image is up-scaled to give a natural look HR image. It can be observed from the result that the proposed method performs better both subjectively and objectively.