The invention provides a vision-guided trajectory planning method on the basis of a Delta robot joint space. The method is applied to the optimal time motion of a Delta robot. The method comprises the following steps: solving a kinematic relation of the Delta robot; establishing an inverse kinematical equation for a motion from an end effector to each joint; acquiring a target position through an industrial intelligent camera; updating the target position on real time through an encoder; dividing a working area of the Delta robot into 9*13 subareas; planning the motion route of each joint at different areas through a B spline offline; maintaining smooth and continuous speed, acceleration and jerk; reducing the impact of a servo motor to a mechanical structure; improving the typical particle swarm optimization; speeding up to search time nodes with optimal solution by the fractional order particle swarm optimization in order to avoid local optimal solution; then performing the two-dimensional fuzzy method to select the optimal time node corresponding to the working area of the robot on line, and thus finishing control.