Smoothing and optimizing the look of images

Images can be smoothed to remove noise before further processing like pixelwise ratio calculation, thresholding or segmentation, or to improve the appearance for presentation purposes. Smoothing can be performed in time and in space.
For printing/publishing the intensity scaling of the images (the relationship between the intensity values and the actual color/brightness as the image appears) has to be optimally set using image scaling and gamma correction to highlight details of interests without distorting data (e.g. clipping other important details).

Smoothing using spatial filtering

Spatial filters filter each frame of the time lapse is performed independently, i.e. act in space:

• When intensity information is (e.g. for ratio calculation) it is important to use linear filters.
• For morphological measurements, however, nonlinear filters, like the Wiener filter is more suitable, because it preserves edges better than linear and median filters.
• Median filtering can be used to remove spotty noise e.g. originating for aggregated/precipitated fluorescence dye in the sample.

Linear smoothing filters:

• Filters/2D Savitzky-Golay filter : performs polynomial smoothing by kernel filtering. Rely on this function when working with intensities.
• Filters/2D DFT Filters... : low pass filters have smoothing effect.

Nonlinear smoothing filters:

• Filters/Wiener filter - the use of Wiener filter is advised before thresholding or segmentation because this filter provides strong noise suppression while smudges sharp edges only little, and faster than the Anisotropic diffusion filter below.
• Filters/Anisotropic diffusion filter
• Filters/2D Median filter

Smoothing using temporal filtering

Temporal filters act over each pixel, spatially independently, filtering the time lapse of intensity values:

• Filters/Temporal Average filter
• Filters/Temporal Median filter
• Filters/Temporal Block filter
• Filters/Temporal Savitzky-Golay filter
• Editing/T or Z-project : as an extreme case the the mean or median of the whole time lapse can be calculated

Intensity Scaling images

How to set image scaling:

• Select Look Up Table (LUT) in the context menu of the Image Window. Gamma values can be entered here.
• Use the Intensity scale bar in the bottom of the Image Window
• Use the Set Scaling dialog in the context menu of the Image Window
• Use the Appearance/Set Scaling/LUT to quickly set the same LUT and scaling in multiple image windows

Rules of scaling images for presentation

• Set the minimum of the scale in order to prevent complete clipping the background. Some background has to be shown in the images to illustrate that no dim details were lost in the presentation.
• Use increased gamma value to visualize dim details without saturating bright objects. Use the same gamma value for all images in a presentation. For publication, some journals require indication of gamma value if other then 1.
• Try to avoid saturation by setting the maximum of the scale properly, however it is not a problem, if some less interesting details are saturated in order to show details of interest.
• When saving movies, often frame-by-frame scaling at percentiles is useful, because the overall brightness of the image stays constant.