There are several papers about pseudo dynamic methods used in signature authentication. Recently, the gray scale features local binary pattern(LBP) originate from texture analysis has been widely used in signature verification system with advantage of robustness to illumination change. The major problem of LBP is its sensitivity to noise, hence many solutions has been applied to solve this problem. In this paper, we further study the performance of LBP in terms of different blocks, then Local directional pattern is explored to obtain a stable and effective feature with the same blocks as LBP. The experiments done with GPDS960Graysignature database demonstrate the effectiveness of LBP and LDP, LBP performs a little better than LDP while LBP has higher dimensions than LDP while the classifier is deployed by Linear Support Vector Machines (SVMs).