IR Image Analysis and Interpretation
Reading patterns, not pixels
Interpretation begins with pattern recognition: shape, edges, gradients and repetition. A thermal bridge usually shows sharp, geometry-following edges; moisture shows broad, diffuse, blotchy areas; a failing bearing shows a compact footprint at the housing; a convection plume shows a soft rising column. Distinguish qualitative work, where relative differences and patterns suffice, from quantitative work, where emissivity correction and calibrated analysis are required.
Tools of analysis
Level and span, isotherms, spot and box measurements, line profiles, histograms and delta temperature tools help isolate features. Freeze and store radiometric files so parameters can be adjusted later; a saved screenshot without radiometric data cannot be re-analysed.
Artefacts and traps
- Reflections of hot or cold objects on shiny surfaces, often mistaken for real defects.
- Emissivity variation across one surface, such as a dirty patch on metal.
- Solar loading, wind cooling, rain or recent occupancy changing surface temperature.
- Judging from one hot pixel instead of a pattern and a comparable reference.
- Inspecting too late or too early relative to the thermal event, especially for moisture and building envelopes.
- Ignoring that a low delta temperature may reflect timing rather than absence of a defect.
Worked example
A flat roof is surveyed for trapped moisture. Because wet insulation has greater thermal capacity, it stays warmer after sunset while dry areas cool faster. The rule of thumb is to wait until a useful delta of at least 2 to 3 K develops across the assembly, typically a few hours after sunset following a sunny day, and to confirm with a second survey or a moisture meter rather than a single image.