In order to improve the performance of cursor selection task in complex target environment under small target cluster interface, an optimization algorithm for fast cursor positioning based on Kalman filter under small target cluster interface is proposed. The influence of factors such as target size, target initial distance, target density and the like on the target selection efficiency and accuracy is comprehensively considered, and the reliability between an observed value and an estimated value of a cursor is calculated in real time by estimating, predicting and correcting the position information and the motion information of the cursor to dynamically predict the motion track of the cursor and implicitly improve the motion stability of the cursor. The probability distribution of potential targets is updated based on Bayesian estimation, and controlled experiments are conducted under different conditions. The results show that the dynamic cursor fast positioning optimization algorithm based on Kalman filter is helpful to achieve accurate target selection and fast interaction.