Fig. 4: Averaging and extraction of the temperature sensitivity and resolution of the three sensors.

a Mean temperature sensitivity and resolution per pixel (1 pixel), 3 × 3 pixels, and 405 typical pixels for the different sensors. Dashed line: typical resolution of infrared (IR) cameras used in medical thermometry (see Results: Experimental design). b ROI is defined as 9 × 90 pixels (light blue rectangle with 9 × 18 pixels in the figure for simplicity). A subsample (pink square) contains 3 × 3 pixels. The 9 waveforms in each subsample are transformed into spectra as previously described in Fig. 3 and then averaged. The initial ROI (9 × 18 pixel matrix) is thus reduced by averaging to a 3 × 6 subsample matrix. c Process for extraction of the sensitivity and resolution, and the standard deviation in one-pixel analysis \((\bar{r},\bar{s})\) and in subsample analysis \((\bar{R},\bar{S})\). For each row of the pixel or subsample matrix defined by the ROI, the minimum (maximum) value of the resolution (sensitivity) is extracted, and the mean and standard deviation are computed (Fig. 4a, first and second bars)