Lanczos algorithm based L-curve and curvature in image resolution enhancement

In this paper, we propose the application of the L-curve and curvature to the super resolution image restoration problem. L-curve criterion is used to estimate regularization parameters. However, the computational of the L-curve is quiet costly for large size problems. The Lanczos algorithm and Gauss quadrature are used to reduce the computational cost. But the many iterations of Lancros algorithm results in the expensive cost of L-curve. The paper develops an approximate technique that can be used to get the proper the iterative times of Lancros algorithm with L-curvature. The technique shows that L-curve and curvature can be computed fairly inexpensively by partial Lanczos algorithm. Then a suitable value of the regularization parameter is determined from L-curve.