Human motion tracking method based on space embedded extreme learning machine

The invention discloses a human motion tracking method based on a space embedded extreme learning machine, mainly aiming to solve the problem of great tracking error in the prior art. The method comprises the following implementation steps: (1) preprocessing a training video and a testing video to obtain a training sample and a testing sample; (2) extracting the image characteristic matrixes of the training sample and the testing sample by using a descriptor, and constructing a combined matrix by using the characteristic matrix of the training sample and a human motion posture matrix; (3) calculating an output weight according to the projection of the combined matrix in a random characteristic space and the human motion posture matrix; (4) evaluating the posture estimation matrix of the testing sample according to the output weight; (5) calculating a difference value between the posture estimation matrix of the testing sample and the human motion posture matrix for serving as a final human tracking error. By adopting the human motion tracking method, the tracking error can be reduced effectively. The method can be applied to motion capture, human-machine interaction and video monitoring.