The inverse of liquid-level based on the surface temperature of a tank is a new liquid-level recognition method with high identification accuracy and wide application range. A physical model for a storage tank is established. The surface temperature field of the tank is derived by a finite volume method and the liquid-level and temperature inside the tank are inverted. In order to verify the recognition accuracy, an experiment for pressure vessel comparison is designed. The experimental results show that the simulation recognition error is less than 1% in the range of acceptable measurement error (σ=2℃) and the experimental recognition level error is about 2%. The liquid temperature inside the cylinder can be recognized by the inversion algorithm and the recognition precision is higher. The experimental result shows that this recognition method is effective and can provide the theoretical and experimental basis for infrared liquid-level detection and temperature recognition.