An improved algorithm for trajectory detection of 3D printer based on three frame difference was proposed to solve the problems of unclear trajectory acquisition and unsatisfactory trajectory detection. USB camera was used to collect the image, then the improved three frame difference method and Gaussian mixture model were used to roughly judge the trajectory. Qt framework was used to detect the trajectory and calculate the velocity of the 3D printer nozzle. The results show that using the algorithm, the trajectory of 3D printer head can be obtained completely and the amount of image processing is reduced to a great extent.