Skip to content

Blur

The Blur command applies a box (mean) filter to the image. Every pixel is replaced by the average intensity of its surrounding neighbourhood.

Because every pixel in the window contributes equally, a box blur is the cheapest possible smoothing operation - just a running average - which makes it fast, but it can also leave faint rectangular “ringing” artifacts around sharp edges that a distance-weighted kernel like Gaussian Blur avoids.

A noisy image next to its box-blurred version

Use Blur as a fast, lightweight noise-reduction step before segmentation. For better edge preservation, consider Gaussian Blur instead.

ParameterDescription
Kernel sizeSide length of the averaging kernel in pixels (must be odd; range 3–27)
  • 3–5 - subtle smoothing, preserves fine structure
  • 7–11 - moderate noise reduction for typical fluorescence images
  • 13+ - heavy smoothing; fine structures may be lost

In image processing, a kernel (convolution matrix) is a small matrix applied to each pixel and its neighbours. A box blur kernel fills every element with 1/N, replacing each pixel with the neighbourhood mean.