itk-GradientMagnitudeImage
The magnitude of the image gradient is used to help determine object contours
and the separation of homogeneous regions. This filter computes the magnitude of the gradient at each pixel, using a finite difference approach.
Usage: itk-GradientMagnitudeImage inputImageFile outputImageFile dimension
Below is a picture of the image, and then the filtered image:
itk-GradientMagnitudeRecursiveGaussianImage
In practice it is convenient to define a scale in which to find the gradient at each pixel. This is done by preprocessing with a smoothing filter. A Gaussian kernel is used with this filter, and an associated scale comes out the value for the standard deviation of the Gaussian.
Usage: itk-GradientMagnitudeImage inputImageFile outputImageFile dimension
Below is a picture of the image, and then the filtered image:
A comparison of the itk-GradientMagnitudeImage and the itk-GradientMagnitudeRecursiveGaussianImage is shown below, in clips of the corners of each filtered image above, quadrupled in size to show the differences:
itk-DerivativeImage
This filter is used to compute the partial derivative of an image along a particular axial direction. The user enters parameters for the order of differentiation and direction:
Usage: itk-DerivativeImage inputImageFile outputImageFile dimension order direction:
Below is a picture of the image, and then the image filtered in the y direction, first order derivative, then second order derivative:
|