<p>Writer verification, a binary problem determining whether two handwriting samples originate from the same person, remains a less explored task despite extensive research on writer identification methods. In writer verification, the ability to compare two given samples for similarity without retraining the model distinguishes it from writer identification, where model retraining is typically required when introducing a new author. This distinction underscores the importance of writer verification in various fields, including forensic analysis, historical document analysis, student authentication, and access control systems. Existing approaches to writer verification have shown promising results, yet there remains significant potential for improvement in terms of execution complexity and verification accuracy, particularly in pinpointing small and isolated discrepancies between handwriting samples. Moreover, these methods often suffer from high computational complexity and parameter utilization. To address these challenges, this paper proposes a novel approach named “LightSiameResNet” for writer verification. This approach integrates a lightweight Siamese network with residual network (ResNet) and Laplacian embeddings, reflecting a streamlined and efficient solution. Specifically, LightSiameResNet leverages ResNet and Laplacian embeddings to enhance the accuracy of the verification process while maintaining a lightweight architecture. Experimental evaluations conducted on benchmark datasets demonstrate competitive performance, achieving high accuracy rates across different classes of handwriting verification tasks.</p>