JA EN

Fourier Transform

A mathematical technique that decomposes an image into spatial frequency components, where low frequencies represent smooth regions and high frequencies correspond to edges and noise.

The Fourier Transform decomposes an image's spatial pixel variations into frequency components. By converting to a frequency-domain representation, it enables processing that would be difficult in the spatial domain. Image processing uses the 2D Discrete Fourier Transform (2D DFT).

Frequency-domain filtering includes:

Computation uses the FFT algorithm at O(N² log N) complexity via np.fft.fft2() or cv2.dft(). The convolution theorem states that spatial convolution equals element-wise multiplication in the frequency domain, enabling speedup for large kernel operations.

Related Terms

Related Articles