부산대학교 도서관

This study introduces a novel approach, leveraging memristor-based fixed-point computing, to implement a high-performance Fast Fourier Transform (FFT) algorithm. Memristors enable direct data processing within memory, addressing power inefficiencies and processing bottlenecks associated with conventional architectures. The proposed method emphasizes parallelism through memristor calculations, enhancing computational efficiency in resource-constrained environments. In the image processing and restoration algorithms, the Memristor-based Fixed-point FFT demonstrates both efficacy and accuracy. Despite precision constraints in memristor-based fixed-point FFT, normalization strategies effectively mitigate noise in image transformations.