filters

Linear convolution filters

Gaussian filter

Implements filtering with a \(k\)-order Gaussian derivative as a series of 1D convolutions. Currently supports up to second-order derivatives.

---

source

gaussian_filter

 gaussian_filter (img:torch.Tensor, sigma:float, order:int|list=0,
                  mode:str='reflect', truncate:float=4.0)

*Convolves an image with a Gaussian kernel (or its derivatives).

Inspired by the API of scipy.ndimage.gaussian_filter and the implementation of diplib.Gauss.*

	Type	Default	Details
img	Tensor		The input tensor
sigma	float		Standard deviation for the Gaussian kernel
order	int | list	0	The order of the filter’s derivative along each dim
mode	str	reflect	Padding mode for torch.nn.functional.pad
truncate	float	4.0	Number of standard deviations to sample the filter
Returns	Tensor

Hessian matrix of an image

Compute a symmetric matrix of all second-order partial Gaussian derivatives of an image.

---

source

hessian_eigenvalues

 hessian_eigenvalues (img:torch.Tensor, sigma:float, **kwargs)

---

source

hessian

 hessian (img:torch.Tensor, sigma:float, as_matrix:bool=False, **kwargs)

Compute the Hessian of a 2D or 3D image.

Frangi

Implements the Frangi filter for enhancing tubular structures in 2D and 3D images. Refer to the original paper: Frangi et al., 1998

---

source

frangi

 frangi (image:torch.Tensor, sigma_range:tuple=(1, 10), scale_step:int=2,
         sigmas:list=None, alpha:float=0.5, beta:float=0.5,
         gamma:float=None, eps:float=1e-10, device:str|torch.device=None)

	Type	Default	Details
image	Tensor		The intput image
sigma_range	tuple	(1, 10)	The range of sigmas to use
scale_step	int	2	The step between sigmas
sigmas	list	None	Optional list of sigmas to use
alpha	float	0.5	Plate-like and line-like structures threshold
beta	float	0.5	Blob-like structures threshold
gamma	float	None	Second-order structure threshold
eps	float	1e-10
device	str | torch.device	None
Returns	tensor