1. Institute of Mathematics, Henan University, Kaifeng, 475004, China 2. College of Mathematics and Information Science, Henan University, Kaifeng, 475004, China 3. Department of Mathematics, University of Bergen, 5009, Bergen, Norway
Abstract:
We apply the dual algorithm of Chambolle for the minimization of the LLT model. A convergence theorem is given for the proposed algorithm. The algorithm overcomes the numerical difficulties related to the non-differentiability of the LLT model. The dual algorithm is faster than the original gradient descent algorithm. Numerical experiments are supplied to demonstrate the efficiency of the algorithm.