Universal approximated real-valued Fast Fourier Transform for image blur detection
Published in 36th Irish Signals and Systems Conference (ISSC), 2025
This paper presents a primary study of a universal approximation approach for the real-valued Fast Fourier Transform (FFT) using an over-trimmed artificial neural network (ANN) for blur detection. By reducing the number of neurons in the hidden layer of the ANN (no activation functions or bias) and dividing the image into sub-blocks, the computational complexity is reduced. The approximate FFT achieves up to ~45% reduction in complexity compared to an untrimmed ANN while still performing well for blur detection — promising for deployment in resource-constrained edge devices. :contentReference[oaicite:0]{index=0}
Recommended citation: Wang, X. & Wu, Y. (2025). “Universal approximated real-valued Fast Fourier Transform for image blur detection.” In *Proceedings of the 2025 36th Irish Signals and Systems Conference (ISSC)*, IEEE.
Download Paper | Download Slides | Download Bibtex