Signature Verification System Using Machine Learning and Image Processing

This paper gives a strong and automated signature verification device that integrates advanced gadget gaining knowledge of and image processing strategies to beautify protection and accuracy in signature authentication. The machine undergoes a sequence of preprocessing steps, which include grayscale conversion, noise reduction the use of Gaussian blur, adaptive thresholding for binarization, and feature extraction using the HOG method. The extracted capabilities are then applied to train a couple of classification models, inclusive of SVM, Random forest, and XGBoost classifiers, to assess their effectiveness in signature verification. A GUI has been advanced to facilitate seamless consumer interaction, allowing individuals to add and affirm their signatures in real time. The experimental results demonstrate that our proposed method achieves an accuracy exceeding eighty%, making it a feasible solution for at ease and dependable authentication in packages which include economic transactions, file verification, and identification validation.