Skip to content

Dark value

Cristel Chandre edited this page Jul 19, 2024 · 3 revisions

The image processing of polarimetry data includes the subtraction of the dark signal from the signal of the sample. The dark signal is due to the dark noise of the camera and refers to the signal acquired by the camera even when no photons related to the sample are incident on the camera. The dark signal value depends on the specific camera used and the temperature of the latter, as well as the image acquisition parameters (gain of camera, exposure time).

PyPOLAR offers the possibility for the dark signal to be calculated automatically, and uses this approach as the default one to calculate the dark signal. The method proceeds as follows: Each image of the polarized image stack is paved with non-overlapping cells of 20x20 pixels. The mean value of each cell is computed for the first image of the stack (first angle). This determines the cell with the smallest mean value. The average over all angles for this specific cell is the Calculated dark value. This value appears in the Advanced tab as the Calculated dark value.

One can alternatively measure experimentally the dark signal by acquiring an image in absence of a sample and calculating the average signal from multiple regions of the field view. Alternatively, one can acquire an image of the sample and calculate the average signal from multiple regions of the field view that do not include the sample. The user can then apply the measured dark signal value to be used for the analysis in the Advanced tab as the “Used dark value”. To this end, the user needs to activate the “Dark” toggle in the Advanced tab and then type in the measured value. For 1PF measurements with the current camera of the spinning disk setup at the Institut Fresnel, the experimentally measured value is close to 480.

The automatic calculation of the dark signal necessitates the presence of regions devoid of sample, i.e., that the field of view is not fully covered by the sample. If the field of view is entirely filled with the sample (e.g., in presence of dense monolayer of cells, or of large tissues such as muscle or Drosophila embryos) the user cannot use such images for automatic dark signal calculation. In such cases, the user will either have to move to a neighbouring region devoid of sample, or impose manually a dark signal value to be used for the analysis, e.g., 480 in the case of 1PF with the current camera at Institut Fresnel.