Fig. 2
From: Spatio-temporal and -spectral feature maps in photoplethysmography imaging and infrared thermography
Video processing chain. Video segments pass through the three stages: ROI generation, temporal filtering and feature generation to create new feature maps. Firstly, pixels are spatially pooled. Images are then subsampled to a resolution which can be processed. Afterwards, pixel series which correspond to ROIs are processed. Consequently, the subsampled video segments are buffered in a first in, first out (FIFO) stack. Features are generated from short-time Fourier transforms (STFT), power spectra (PS) or signal statistics. An optional preprocessing adds artificial frames if there were losses during recording. We introduce a postprocessing for evaluating the maps: ‘Similarity maps’ are created by using histogram intersections and the root mean square (RMS) contrast is calculated for a manual selection of ROIs. The parametrization used in this work is given in Table 3 and the method is provided in Section "Methods" in detail