Nonlinear structure analysis of handwritten data is important for handwritten data recognition and signature authentication.
However, handwritten data usually has high dimension and dimension reduction processing is required. In this paper, an improved
ISOMAP algorithm is proposed, improving the efficiency of ISOMAP algorithm. It is applied to the nonlinear structure analysis of
handwritten data. The experimental results show that the proposed algorithm has high efficiency and accuracy, and the handwritten data has latent intrinsic dimension and nonlinear structure.