To accurately measure the pose of a parallel machine tool and improve its motion precision,a vision-based pose measurement method for end pose detection in parallel mechanisms was proposed.A measurement platform was built using the Femto depth camera as the measurement device,and the camera was calibrated using a chessboard pattern and MATLAB.A target was placed on the moving platform to collect pose data,and the collected data were filtered and extracted using a point cloud database.Finally,the calculated actual poses were experimentally compared with the input poses.The results show that the maximum errors on the X，Y and Z axis are 0.071 mm,0.047 mm,and 0.394 mm respectively,the maximum errors of rotating around the X,Y and Z axis are 0.09°,0.09° and 0.14° respectively,which meets the accuracy requirements and confirms the effectiveness of the proposed method.The proposed method offers efficient and convenient operation with low cost,and while ensuring high detection accuracy.It can provide new ideas for pose measurement and subsequent calibration experiments.