Bettemir, Onder Halis2024-08-042024-08-0420231302-09002147-9429https://doi.org/10.2339/politeknik.1144858https://hdl.handle.net/11616/102803Vast majority of the existing buildings in Turkiye are not inspected for thermal insulation quality during the construction process therefore, thermal insulation performance of the existing buildings cannot be known. Measuring the thermal insulation performance of the buildings by scraping the plaster and examining the heat insulation material is not a viable solution when the size of the building stock of Turkiye is considered. In this study, detection of thermal bridges of the buildings by processing the thermal images of the buildings is proposed. The method is based on the binarization of the thermal image by the classification of the building elements as heat loss element or no heat loss element by analyzing the thermal image of the building. Global threshold methods and adaptive local threshold methods applied for binarization. All of the implemented methods require a threshold value for the classification. Determining a valid threshold value for all images is not possible therefore the threshold value is determined by the Otsu algorithm. Threshold determination process is executed both on the thermal image and the edge image. Obtained threshold values are implemented on the thermal images and the edge images. Local edge detection algorithms derived from the literature are compared by examining five thermal images and the comparison revealed that the Modified II Frei-Chen and Second-order Laplace operator provided the most suitable result. The case studies revealed that the thermal insulation performance of the existing building stock can be determined quickly, economically and reliably by implementing the proposed method.eninfo:eu-repo/semantics/openAccessEdge detectionbinarizationotsu algorithmArticle10.2339/politeknik.1144858WOS:000957557300001Q4