Due to the increasing use of composite patches for repairing or reinforcing metallic and composite structures, health monitoring of these patches has become critically important. Thermography is one of the most powerful non-destructive testing methods employed for evaluating repaired structures. Given the numerous parameters influencing defect detection in repaired composite structures, identifying the most effective approach for determining these parameters is of great significance. In this study, experimental and finite element simulation methods were used to investigate key parameters for thermographic inspection of carbon fiber composite patches, such as the duration of thermal loading and the optimal imaging frame. The composite patch used to repair the carbon fiber base structure contained delamination and disbonding defects of various sizes and characteristics at different depths. In the thermal simulation, a criterion based on the maximum thermal variation along a path was developed to extract the optimal thermal imaging frames. Validation of the results was performed by defining direct and diagonal paths across the defects in both experimental thermographic images and simulated temperature contours. The average error in thermal variation across different paths in thermographic and simulation images was less than 4.5%, confirming the consistency of results from both methods. Another major challenge in defect detection process is determining the impact of defect characteristics on the detection process and prioritizing these features. In this research, defect attributes—including size, depth, material type, and distance from the edge—were examined using methods developed in the present study. The findings indicate that the influence of defect characteristics on detection follows the order: material type, depth, size, and proximity to free edges. Furthermore, the study revealed that the distance between two defects significantly affects image contrast in thermographic detection. The relationship between defect spacing and defect dimensions was extracted through simulation method and validated using experimental thermographic images.