ITK Pre-Processing Algorithm Wrappers in Python/MeanFilter

Mean Image Filter

This is implemented using the itkMeanImageFilter class in ITK. This takes an image and performs a mean value of local neighbourhood region described by the radius parameter, for each pixel. This filter can be used to reduce the noise present in an image; however, it results in smoothing of the edges of various objects in the image.

Parameters:

• 1. Radius - for an n-D image, this is an n-tuple, with each entry describing the radius of the neighbourhood in that dimension. For example, for a 2-D image, a Radius of (1,1) indicates a neighbour region of size 3x3 around the pixel. Small values of radius using the mean filter can be useful in reducing the Gaussian noise present in the image, but not affecting the image. Large values of radius can reduce the noise level; however, this will be accompanied by increased blurring of objects in the image. Here is an example of the use of the mean filter on seg_orig. Fig. 1 shows the image obtained using a Radius of (1,1) or a 3x3 neighbourhood, and Fig. 2 shows the image obtained using a Radius of (3,3) or a 7x7 neighbourhood. Clearly, mean filtering with a 7x7 neighbourhood smoots the edges of the images significantly.

Back to main page.

Fig.1 Mean filtered image with 3x3 neighbourhood

Fig.2 Mean filtered image with 7x7 neighbourhood