To improve the accuracy of the commonly-used Extended Kalman Filter that omits the high-order terms of the derivation of the system equations, the second-order Adaptive Extended Kalman Filter (AEKF) was furtherly developed and Unscented Kalman Filter (UKF) was introduced based on two-compartment model. The results obtained by first-order AEKF, second-order AEKF and UKF based on numerical simulations and in vivo experiments were assessed and compared. The results demonstrate that the fluorescence pharmacokinetic parameter reconstruction based on first-order AEKF and second-order AEKF are similar, while the parameters obtained based on UKF are optimal in terms of quantitativeness and contrast-to-noise ratio, which is consistent with the theory that UKF should have higher accuracy because it does not ignore higher-order terms, and validate the feasibility and effectiveness of our proposed methods.