Detection of Forged Handwriting Through Analyzation of Handwritten Characters Using Support Vector Machine

People often use a keyboard to input data in digital form. However, there are still some cases where handwriting is still used and often in significant scenarios such as cheques. The current study focuses mainly on detecting forgery in a person's signature or cases where original handwriting was altered or additional characters were added. Thus, the study proposed a handwriting forgery detection system that utilizes image processing and Support Vector Machine (SVM), a linear classification model. The system will take the original handwriting of a person as its training data to create a model that would evaluate whether the presented handwriting is original or forged. In addition, SVM will also be used for text recognition of handwritten letters. The models are then evaluated using a confusion matrix and F1 score. The evaluated result for the text recognition model achieved an F1 score of 0.9052. On the other hand, the forgery detection model had an F1 score of 0.6013.