Image magnification method using Compressed Sensing

In this paper, a new method for image magnification is presented. The image reduction is seen as a result of image multiplying with a compressed matrix, and the magnification is stated as an inverse problem of reduction. We exploit the reconstruction idea of Compressed Sensing and propose a norm minimization model to solve the inverse problem. The norm reflects the image's natural property - compressive in transform domain and local smoothness in space domain, its minimization is the magnification result. The experimental results show that the enlarged images produced by the new method have higher precision and hence offer more detail information than traditional methods.