Detecting features of human personality based on handwriting using learning algorithms

Handwriting analysis is useful for understanding the personality characteristics through the patterns created by the handwriting and can reveal features such as mental and emotional instability. On the other hand, it is difficult to determine the personality, especially when it is associated with the law because there is no threshold or scale being able to make detailed results of the analysis. This thesis aims to provide an automated solution to recognize the personality of the author by combining image processing and pattern recognition techniques. The personality recognition system proposed in this project is composed of two main parts: training and testing. In the training part, after feature extraction from all image patterns of the input text, a proportional output is created through the MMPI personality test. Then these inputs are trained to the neural network as a pattern. As a result of this training, a comprehensive database will be formed. In the testing part, the database is used as a main comparison reference. After feature extraction, the input text image is compared with all patterns in the database to find the closest image to the input text image. Finally, the MMPI personality test output for the proposed text image is introduced as the output personality parameters.