In this paper, a novel method for human pose estimation algorithm is proposed. The proposed algorithm only uses depth information only and can work for any human without calibration. To estimate human poses, the proposed algorithm combines a support v...
In this paper, a novel method for human pose estimation algorithm is proposed. The proposed algorithm only uses depth information only and can work for any human without calibration. To estimate human poses, the proposed algorithm combines a support vector machine(SVM) and a geodesic distance graph. The SVM-based pose estimator uses randomly selected human features to reduce computation. Body parts that involve a lot of motions are estimated by geodesic distance value. The proposed human pose estimation algorithm is evaluated through an experiment and the result showed that our method perform fairly well.