Handwriting stroke reflects the written trace of each individual's rhythm and style. By examining all elements of handwriting and interpreting them separately or integrated, we could generate a sketch of the writer's character traits, emotional disposition and social style using standard of graphology. As image, the analysis of graphology is divided into two approaches that graphics features and segmentation digit each character. In this research, using combination of graphical approach based on signature and digit of character of application form using multi-structure algorithms and artificial neural networks (ANN). The image split into two areas: the signature based on nine features and application form of letters digit area. Each area had pre-processing performed to improve the recognition accuracy. Signature area is classified using ANN based on five features which result an accuracy of 56-78%. While four feature of the signature that detection using multi structure algorithm result 87-100% accuracy. In meantime, pattern recognition of application form digit area using Learning Vector Quantization gave 43% accuracy. It used 100 sets of data testing after training with 10-25 data. The system has been implemented with the software so that it can be used for classification of personality from handwriting scanned automatically.