Non-destructive detection of internal defects in trees using infrared thermography

Abstract

Trees are vital to ecosystems, playing a key role in regulating the natural environment, climate, and urban forestry. However, trees can pose risks in urban areas due to potential property damage from falling branches or entire trees. This is especially true when these trees have internal defects that threaten their stability and health. Monitoring tree health is essential for understanding these risks and making informed decisions about tree management. Internal axial tree cavities, often caused by timber decay or termite attacks, are a primary cause of tree failures. Trees can adapt by reinforcing tissue around these voids, but large cavities can eventually compromise structural integrity. Visual inspections often fail to detect hidden internal defects, and current techniques used for defect detection can be invasive. This study aims to address the problem of detecting internal defects in trees qualitatively using nondestructive infrared thermography. By comparing the surface temperatures of healthy and unhealthy tree trunk sections, this study evaluates the efficacy of the method used, and analytical coding is simplified in the image processing stage. Sap circulation influences the temperature differential along a tree trunk, enabling the differentiation between functional and dysfunctional tissue. Temperature variations on the tree surface may indicate deteriorating tissue, cavities, or defects. Because of this favorable thermal profile, live tree imaging can be performed using the passive method. A total of 45 trees representing 40 species were selected for the study, 38 from the Royal Botanical Gardens and 7 from the University of Peradeniya. The FLUKE TI-105 camera was used to capture both infrared and digital images, and thermograms were processed using MATLAB and Fluke Connect Desktop software. A two-week survey identified defective trees, with detailed records of each tree's common name, scientific name, and diameter at breast height (DBH) while GPS coordinates were used to map tree locations. The defects included cankers, external deterioration, internal cavities, and cracks. Image acquisition was conducted on a low-rainfall day to ensure data integrity, considering unusually heavy rains during the study period. Observations included distance, ambient temperature, and relative humidity, with the camera's emissivity value set to 0.95. The study conclusively demonstrates that some internal defects in trees can be detected using thermograms. Defected areas were typically in lower temperatures than the surrounding areas and capturing images at optimal times is crucial to avoid altered temperature profiles caused by direct sunlight or rain. Moss on the trunk surface was found to cause misleading thermal profiles. Since images were captured from a distance, background masking was essential for effective analysis. Higher-contrast thermal images were beneficial for pinpointing defects. When this was not feasible, k-means clustering was preferred despite its limitation of generalizing temperature profiles, particularly with low-temperature spots in lighter colors. Combining k-means clustering with edge detection improved defect identification. The study emphasized that such analysis is case-specific, as influencing factors vary. Further research is needed, especially in regions like Sri Lanka where local tree species have been minimally studied in this context.