When textural features are applied to handwriting identification, the accuracy is not high due to the fact that the textural features fluctuate randomly because of the different contents and positions of the characters. In this paper an approach for the construction of texture images is presented. The features of the texture images are calculated and the mathematical expectations of them are estimated. Then the weighted Euclidean distance (WED) classifier is adopted to identify the writer. This approach is proved to be a very promising method for text-independent handwriting identification both theoretically and practically.