Figure 1From: Shotgun approaches to gait analysis: insights & limitationsIllustration of template matching procedure. The algorithm starts with an initial estimate of events in a signal (panel a). Around each event an epoch (window) is defined (grey shading in panel a), and the epochs are averaged to define a template (panel b). This template is compared with every epoch using the corresponding cross-correlation function. In the next step, events are shifted by the time lag of maximal cross-correlation (panel c). These events then serve as a new initial estimate (panel a). This procedure is repeated until convergence. Panels d &e show the final events and template.Back to article page