The new coating structure exhibits outstanding properties, including high-temperature strength, excellent fracture toughness, and strong ablation resistance, making it widely applicable in high-temperature environments such as aviation, aerospace, internal combustion engines, and nuclear reactors. However, the complex preparation process and harsh operating conditions of these coatings often lead to issues like uneven thickness, peeling, and internal substrate defects, all of which can compromise equipment control and operational safety. Infrared thermal wave nondestructive testing (NDT) technology, with its non-contact, intuitive, and efficient approach, offers an innovative solution for detecting structural defects in coatings. In this study, we conducted a comprehensive investigation into key aspects of pulsed infrared thermal wave detection for coating structures. This included analyzing and simulating the detection principles, developing a pulsed infrared thermal wave detection system, selecting optimal detection parameters, designing algorithms for processing infrared pulse image sequences, quantifying coating thickness uniformity, and identifying and measuring defect characteristics. We analyzed heat flux transfer in light-pulsed-excited coatings, developed a one-dimensional analytical model of the surface temperature field, and established a clear analytical relationship between coating thickness and temperature variations. Simulations were used to examine how defect geometry and detection parameters influence surface temperature signals, thereby assessing the capability of pulsed infrared thermal wave imaging to detect coating thickness variations and internal defects. A flash pulse excitation infrared thermal wave imaging system was built to facilitate efficient and reliable detection of coating defects. Experimental research further evaluated the influence of defect geometry and detection parameters on surface temperature signals, verified the theoretical model accuracy, and defined a reasonable range of detection parameters. Various pulsed infrared image sequence processing algorithms—including polynomial fitting time differentiation-correlation coefficient, pulse phase method (PPT), and Markov-principal component analysis (Markov-PCA)—were tested to process thermal sequences, extract temperature signal features, and enhance image quality. Notably, the proposed Markov-PCA method significantly increased the signal-to-noise ratio, improving defect detection capabilities. Additionally, we investigated a quantitative method for detecting coating thickness variations and addressed the thermal conductivity inversion problem for coating thickness distribution based on surface normalized temperature distribution. The simulated annealing algorithm enabled accurate thickness inversion calculations, achieving coating thickness measurements in the range of 46–113 μm with a maximum measurement error of less than 10% compared to conventional eddy-current scanning results.