Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{206643,
        author = {Dr. Annapurna H},
        title = {A Writer Stability-Based Approach for Offline Signature Verification},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {2},
        pages = {2173-2180},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=206643},
        abstract = {Offline signature verification is a widely used biometric authentication technique; however, its performance is often affected by intra-writer variability and skilled forgeries. This paper proposes a Writer Stability-Based Approach for Offline Signature Verification, which exploits the consistency of writer-specific signature characteristics to improve verification reliability. Initially, signature images are preprocessed through grayscale conversion, median filtering, Otsu's thresholding, noise removal, cropping, and normalization. Deep feature vectors are then extracted using a pre-trained ResNet18 network. A Writer Stability Model (WSM) is constructed for each writer by estimating the median feature prototype and the corresponding feature stability using the median absolute deviation. Stability-based weights are assigned to emphasize consistent features during matching. The similarity between a questioned signature and the writer's stability model is computed using weighted cosine similarity, followed by threshold-based decision making. The proposed method is evaluated on the CEDAR offline signature dataset using four training-testing split ratios (40:60, 50:50, 60:40, and 70:30). Experimental results demonstrate that the proposed approach achieves the highest verification accuracy of 80.06% with the 40:60 split. The results indicate that incorporating writer-specific feature stability provides a simple, effective, and computationally efficient solution for offline signature verification.},
        keywords = {Offline signature verification, writer stability model, ResNet18, deep feature extraction, weighted cosine similarity, biometric authentication.},
        month = {July},
        }