Multi Noise Classification in Images Using Fast-Fourier Transform and Power Spectrum Density

Noise in data is an enormous barrier to the performance of classification algorithms in a number of real-world circumstances. When multiple sources of noise concurrently impact data, traditional classification techniques such as linear classifiers or simple decision trees often struggle to accurately identify the noise. We present a novel method for multi-noise classification in this work. By using well-known signal processing methods—Fast Fourier Transform (FFT), and Power Spectral Density (PSD) analysis we offer a thorough method for multi-noise classification. The suggested methodology first preprocesses noisy signals to extract significant frequency-domain information. Multiple evaluations are carried out utilizing different benchmark datasets comprising a variety of noise types, such as Gaussian noise, impulse noise, motion noise, and mixtures of these noises, in order to assess the performance of the suggested approach.