Localized infection of a plant can be mapped by a sequence of images capturing chlorophyll fluorescence transients in actinic light. Choice of the actinic light protocol co-determines fluorescence contrast between infected leaf segment and surrounding healthy tissue. Frequently, biology cannot predict with which irradiance protocol, in which fluorescence image of the sequence, and in which segment of the image there will be the highest contrast between the healthy and infected tissue. Here, we introduce a new technique that can be applied to identify the combination of chlorophyll fluorescence images yielding the highest contrast. The sets of the most contrasting images vary throughout the progress of the infection. Such specific image sets, stress-revealing fluorescence signatures, can be found for the initial and late phases of the infection. Using these signatures, images can be divided into segments that show tissue in different infection phases. We demonstrate the capacity of the algorithm in an investigation of infection of the model plant Arabidopsis thaliana by the bacterium Pseudomonas syringae. We show that the highest contrast is found with transients elicited by fluctuating, harmonically modulated irradiance with long periods.