A computer implemented data processing method for recursively approximating a proper value for a target matrix includes the following steps of: determining whether the target matrix corresponds to a low complexity condition; if so, obtaining a first updated target matrix according to a first variance, relevant to a second iteration parameter, and a first iteration parameter, wherein the first and the second iteration parameters correspond to fixed values; if not, obtaining a second updated target matrix according to a second variance, relevant to a fourth iteration parameter, and a third iteration parameter, wherein the third and the fourth parameters are related to the target matrix.