Classification of Handwriting Margin Patterns Using Ensemble Bagging Decision Tree

Abstract

Analysis of handwriting margins plays an important role in graphology, as margin patterns are often associated with individual behavioral tendencies and personality traits. Therefore, detecting and classifying margin characteristics is essential to support automated handwriting analysis using computational approaches. This study uses a computer vision-based approach to classify left-margin handwriting patterns into widening and narrowing categories. The classification is performed by analyzing margin characteristics extracted from scanned handwriting images. The processing pipeline consists of image preprocessing, hybrid feature extraction, and classification using an Ensemble Bagging Decision Tree model. The preprocessing stage enhances image quality through grayscale conversion, contrast adjustment, adaptive thresholding, and noise removal, followed by Region of Interest extraction to focus on the handwriting area. The feature extraction stage applies a hybrid strategy that combines line-based margin analysis and global spatial features to capture both local variations and overall structure. Model performance is evaluated using stratified 5-fold cross-validation to ensure reliable and unbiased results. The experimental findings show that the method achieves an average accuracy of 84.91%, with relatively balanced precision, recall, and F1-score across both classes. These results indicate that margin-based features are effective for representing handwriting patterns in classification tasks. However, variations in writing style and noise from the scanning process may influence performance. Overall, this study demonstrates that the applied approach provides reliable classification results and has potential for further improvement through feature expansion and more advanced learning models.