This study presents the estimation of crack depth by analyzing temperatures extracted from thermal images and environmental parameters such as air temperature, air humidity, illumination. The statistics of all acquired features and the correlation coefficient among thermal images and environmental parameters are presented. The concrete crack depths were predicted by four different machine learning models: Multi-Layer Perceptron (MLP), Random Forest (RF), Gradient Boosting (GB), and AdaBoost (AB). The machine learning algorithms are validated by the coefficient of determination, accuracy, and Mean Absolute Percentage Error (MAPE). The AB model had a great performance among the four models due to the non-linearity of features and weak learner aggregation with weights on misclassified data. The maximum depth 11 of the base estimator in the AB model is efficient with high performance with 97.6% of accuracy and 0.07% of MAPE. Feature importances, permutation importance, and partial dependence are analyzed in the AB model. The results show that the marginal effect of air humidity, crack depth, and crack temperature in order is higher than that of the others.

IR camera has been used widely for the temperature measurement and fault detection of the moving bodies and rotating bodies. The high-speed performance of the IR camera and a reliable thermal analysis method are required for the condition monitoring of the railway vehicle running at high speed. The effective fault detection method using a thermal image analysis could make a real time monitoring of the high speed train possible. Therefore the investigation of the performance of the thermal image analysis method was performed to find the effective thermal image data analysis method. The results suggested that the comparison of the characteristics of the temperatures obtained at different conditions and a continuous temperature subtraction method could be used as a useful analysis method for detecting abnormal temperature condition and histogram equalization could also help to enhance the fault detectability by increasing the contrast of the thermal image

In this study, an analysis were conducted to utilize the thermal infrared image using drone to present the temperature correction method of thermal infrared image and the thermal environment by the type of land cladding. The analysis was applied to the temperature correction of the thermal infrared image and total eight thermal infrared images were produced based on the land surface temperature. The thermal infrared image compared accuracy through RMSE calculation. Based on the result of RMSE, the thermal infrared image corrected by the land surface temperature was relatively accurate and contained at 2.26 to 3.58. According to the results, it is expected that the aggregation and waters will perform the functions of the green park sufficiently to improve the thermal comfort and improve the microclimate stability using the thermal infrared image and the reclassified land cover map. The results of this study obtained by Drone and the usability of the drone thermal infrared image in the detection of the thermal environment. Finally, it is expected to contribute to the improvement and management of the thermal environment in the city by being used as a basic data for the improvement and management policy of the thermal environment. Moreover, the macro view is expected to contribute to the mitigation of urban temperature reduction and heat island.