Image Interpolation with Functional Coefficient Local Polynomial Regression
Liyun Su, Qian Yang · International Journal of Applied Mathematics & Statistics/International journal of applied mathematics and statistics · 2013
This paper investigates a novel method for digital image interpolation with functional coefficient local polynomial regression(FCLPR). The orders of local polynomials is discussed and select the optimal order for functional-coefficient that maximizes peak signal to noise ratio (PSNR) based on high resolution digital images. The proposed digital image superresolution reconstruction scheme with nonparametric regression model is compared it with the traditional interpolation methods including nearest neighbor replacement (NNR) and bilinear interpolation(BLI). Interpolation results for real digital images show that the presented method is better than the two traditional one in the sense of improving PSNR of high resolution image and is also superior to the two traditional interpolation methods in dealing with additive noise for low-resolution images. Therefore, the FCLPR method for digital images interpolation from low-resolution images is feasible and effective.