Infrared thermography non-destructive testing technology has been widely used in the detection of composite structures due to its advantages, including non-contact operation, rapidity, low cost, and high precision. To address the challenge of surface crack detection in composite materials, this study proposed a non-destructive testing method based on laser-line thermography and established a crack depth characterization model based on the maximum surface temperature gradient difference. Through theoretical analysis of heat conduction, the mechanism by which surface cracks disturb the temperature field of the linear heat source was thoroughly investigated. Then, combined with finite element numerical simulations, a quantitative relationship model between crack depth and the maximum temperature gradient difference was developed. Finally, the effectiveness of this method for identifying surface crack depth in fiber-reinforced composites was validated through experiments. The experimental results validated that this method achieves high measurement accuracy, with the relative error of crack depth identification below 5%. The crack depth characterization method based on the maximum temperature gradient difference proposed in this study offers a reliable approach for the quantitative detection of surface cracks in engineering structures, demonstrating significant practical application value.