Human intention recognition plays an important role in human-machine collaboration. In this paper, the intention recognition method for target capture is studied, a vision based method is used in the human-computer interaction, and the trajectory of the body's upper extremity is taken as the main criterion. The user interaction intention model is established by using the multi-dimension judging factor. Then the target fast judging algorithm is optimised by adopted the D-S theory to realize the intention factor fusion. Experiment results show that the intention-identification method proposed in this paper can quickly and accurately identify the operator's grasping intentions and serve human-computer collaboration. The comparison experiments show that the multi-dimensional factor intention identification method is more accurate and more effective than the single factor method.