imgcv.filters.freq_domain

Functions

`fft2`(image)	Calculates 2D Fast Fourier Transform of an image. It is a recursive function based on the Cooley-Tukey algorithm. Computes the 2D DFT of an image or array of images in O(n^2 log n) time.
`ifft2`(specturm)	Calculates 2D Inverse Fast Fourier Transform of a specturm. It is a recursive function based on the Cooley-Tukey algorithm. Computes the 2D IDFT of a specturm or array of specturms in O(n^2 log n) time.
`fftshift`(spectrum)	Shifts the zero frequency component to the center of the spectrum.
`low_pass_filter`(img, cutoff[, order, type, return_img_fft])	Apply low pass filter in the frequency domain to an image.
`high_pass_filter`(img, cutoff[, order, type, ...])	Apply high pass filter in the frequency domain to an image.

Module Contents

imgcv.filters.freq_domain.fft2(image)

Calculates 2D Fast Fourier Transform of an image. It is a recursive function based on the Cooley-Tukey algorithm. Computes the 2D DFT of an image or array of images in O(n^2 log n) time.

NOTE: Currently works only for square images and gray scale with size as power of 2.

Parameters:: image (np.ndarray) – Input image to calculate 2D FFT.
Returns:: 2D FFT of the input image. The output is a complex array.
Return type:: np.ndarray

imgcv.filters.freq_domain.ifft2(specturm)

Calculates 2D Inverse Fast Fourier Transform of a specturm. It is a recursive function based on the Cooley-Tukey algorithm. Computes the 2D IDFT of a specturm or array of specturms in O(n^2 log n) time.

Parameters:: specturm (np.ndarray) – Input specturm to calculate 2D IFFT. The input is a complex array.
Returns:: 2D IFFT of the input specturm. The output is a complex array.
Return type:: np.ndarray

imgcv.filters.freq_domain.fftshift(spectrum)

Shifts the zero frequency component to the center of the spectrum.

Parameters:

spectrum (np.ndarray) – Input spectrum to shift.

Raises:

ValueError – If spectrum is not a numpy array.
ValueError – If spectrum is not 2D.
ValueError – If spectrum is not square and size is not power of 2.

Returns:

Shifted spectrum.

Return type:

np.ndarray

imgcv.filters.freq_domain.low_pass_filter(img, cutoff, order=None, type='butterworth', return_img_fft=False)

Apply low pass filter in the frequency domain to an image. NOTE: Currently works only for square images and gray scale with size as power of 2.

Parameters:

img (np.ndarray) – Input image to apply low pass filter.
cutoff (int) – Cutoff frequency of the filter.
order (int, optional) – Order of the Butterworth filter. Defaults to None. If None, order is set to 1.
type (str, optional) – Type of the filter. Choose one of ‘ideal’, ‘butterworth’, ‘gaussian’. Defaults to “butterworth”.
return_img_fft (bool, optional) – If True, return the filtered image in the frequency domain along with the original image in the frequency domain. Defaults to False.

Raises:

ValueError – If input image is not a numpy array.
ValueError – If input image is not 2D.
ValueError – If cutoff frequency is not an integer.
ValueError – If order is not an integer.
ValueError – If filter type is invalid.

Returns:

Filtered image in the spatial domain. (np.ndarray, np.ndarray): Filtered image in the frequency domain and the original image in the frequency domain (if return_img_fft is True)

Return type:

np.ndarray

imgcv.filters.freq_domain.high_pass_filter(img, cutoff, order=None, type='butterworth', return_img_fft=False)

Apply high pass filter in the frequency domain to an image. NOTE: Currently works only for square images and gray scale with size as power of 2.

Parameters:

img (np.ndarray) – Input image to apply high pass filter.
cutoff (int) – Cutoff frequency of the filter.
order (int, optional) – Order of the Butterworth filter. Defaults to None. If None, order is set to 1.
type (str, optional) – Type of the filter. Choose one of ‘ideal’, ‘butterworth’, ‘gaussian’. Defaults to “butterworth”.
return_img_fft (bool, optional) – If True, return the filtered image in the frequency domain along with the original image in the frequency domain. Defaults to False.

Raises:

ValueError – If input image is not a numpy array.
ValueError – If input image is not 2D.
ValueError – If cutoff frequency is not an integer.
ValueError – If order is not an integer.
ValueError – If filter type is invalid.

Returns:

Filtered image in the spatial domain. (np.ndarray, np.ndarray): Filtered image in the frequency domain and the original image in the frequency domain (if return_img_fft is True)

Return type:

np.ndarray