CUDA-Based Click Point Password Authentication System

Abstract

Authentication remains a critical aspect of information security, and while traditional text-based passwords dominate, they suffer from weak memorability and vulnerability to common attacks. Graphical password schemes, particularly click-point systems where users select specific points on images, offer improved usability and resistance to certain text-based threats. However, these methods face challenges such as predictable hotspot selection and increased computational demands when employing mitigation techniques like randomized image transformations and persuasive cued click points. To address this, a CUDA-based click-point password authentication system is proposed, leveraging GPU parallelism to accelerate image rendering, hotspot randomization, and password validation in real time. By offloading intensive computations to GPU kernels, the system achieves enhanced security while maintaining sub-second response times, ensuring both usability and scalability. This integration demonstrates that GPU-accelerated graphical authentication can effectively balance security, performance, and user experience, offering a robust pathway for future secure authentication frameworks.