A Fast Edge-based Image Interpolation Algorithm

In the field of image processing, image interpolation technology is a basic andcommon image processing technology, which can construct a high-resolution images froma corresponding low-resolution image. As the aspects of image video transmission,standard-definition video to high-definition video, color image restoration, target detectionand recognition, medical image analysis, satellite remote sensing image analysis, militaryradar image analysis, and astronomical image analysis, image interpolation has importantapplications. Classic nearest neighbor interpolation, bilinear interpolation, and bicubicinterpolation are non-adaptive interpolation algorithms, which have been used in somecommercial hardware and software of image processing. However, these algorithms usedin enlarging the image interpolation, generally result in blurring and aliasing in the edgeregion. Adaptive interpolation algorithm can mitigate this blurring and aliasing to a certainextent. However, the computational complexity of this kind of algorithms increaseexponentially, which will limite the applications.In order to reduce the blurring and aliasing, improve the visual quality of theinterpolated image, and have low computational complexity, we propose a fast imageinterpolation algorithm based on edge information of images. The algorithm comprisesthree steps: non-edge and the edge pixels division, bilinear interpolation on the non-edgepixels, and adaptive interpolation on the edge pixels. We have made experiment with grayimages and color images, and analyse the results from the nearest neighbor interpolationalgorithm, bilinear interpolation algorithm, NEDI algorithm, FEOI algorithm, and theproposed algorithm. After comparing the visual quality of interpolated images and thetime cost, we can make the conclusion that the proposed algorithm can effectively reducethe aliasing and blurring to have better interpolated image, and reduce the computationalcomplexity to make the proposed algorithm having important application in real-timeimage processing.