A Dai-Kou-Type method with image de-blurring application

By exploiting nice attributes of three-term conjugate gradient (TTCG) methods and the efficiency of the Dai-Kou scheme, this paper proposes a spectral class of Dai-Kou-type methods for monotone systems with convex constraints. The scheme combines a modified Dai-Kou search direction with the projection method and can best be described as an adaptation of the Dai-Kou method to nonlinear systems. An important contribution of the scheme is its application to image de-blurring. The method is shown to converge globally under mild assumptions. Furthermore, test results of some numerical experiments suggest that the proposed approach outperforms three recent schemes for convex constrained monotone systems.