The improved LevenbergMarquardt algorithm was used to estimate the parameters of poplar branch and trunk modeling. To achieve a dynamic three-dimensional visualization of poplar, amount of diameters were obtained by measuring different poplars forest age. According to the growth trend of diameters, the function of diameter growth was built using the improved LevenbergMarquardt algorithm. A class of heart-shaped curve equation was proposed, and the relationship function between the main branch length and the depth of the sticks was obtained using the improved Levenberg-Marquardt algorithm. A lot of comparison experiments were carried out, and the results demonstrated that the fitting algorithms and mathematical models selected could better simulate the growth of poplar branches. The three-dimensional visualization of poplar was realized with dynamic simulation equation, and the results validated the effectiveness of poplar growth modeling.