Local Polynomial In The Application Of Image Processing

Local polynomial regression is one of the three important methods in nonparametricregression and modeling. In the last two decades, the basic theory of local polynomialregression was extensively developed. And it has been widely used in nonlinear time series,communication, image processing, finance and other fields. The research on application oflocal polynomial has developed rapidly abroad, and it has also been paid more and moreattention in our country.In this paper, we firstly introduce the main topics and the relevant background, anddescribe the related theories of local polynomial in detail. Then a bivariate localpolynomial model is established for image processing, and its application in imageprocessing is investigated as well. The processing results show that our method has a goodeffect. The main contents of this paper are arranged as follows:1. Firstly, we establish a bivariate local polynomial model, and use MATLABprogram to interpolate an image.2. Further, the peak signal to noise ratio of the interpolated image is computed, andwe select the proper order and the optimal bandwidth at the maximum peak signal to noiseratio.3. In the end, we compare the local polynomial interpolation method with thetraditional ones in both cases of noise image interpolation and noise-free imageinterpolation. In the case of noise-free image, local polynomial interpolation method isbetter than the traditional methods, such as nearest neighborhood method, bilinearinterpolation and cubic convolution interpolation. In the processing of noise image, localpolynomial interpolation method is also superior to the nearest neighborhood method andbilinear interpolation method, and as effective as cubic convolution interpolation. All ofthose indicate that the local polynomial modeling is feasible and effective for imageprocessing.