Strictly speaking

image magnification (image zoom or image resample ) is a morbid problem. Based on various image models

several methods such as bilinear interpolation

bicubic spline interpolation

fractal based interpolation and wavelet based interpolation have been developed. The focus problem of image magnification is how to get good visual resolution. In the methods based on wavelet transformation

it is how to make the detail coefficients of the magnified images. The present method interpolates the coefficients straight in the transformed domain

but the experiments show that the magnified images have bad visual effect. Experiments on one dimension signals in this paper illustrate that method is not reasonable. In addition

two other questions are also discussed: how to estimate the method of image magnification and how to choose the appropriate wavelets in the wavelet based method. Comparing and analyzing show that the best means to estimate the method of image magnification is subjective measure

on the other hand the objective measures mentioned in this paper are merely fit for the traditional interpolation methods. Since there is not a perfect method constructing the detail coefficients of the magnified images at present

how to choose the appropriate wavelets is a new question that should be deeply researched.