Motion history of skeletal volumes and temporal change in bounding volume fusion for human action recognition

摘要: Human action recognition is an important area of research in computer vision. Its applications include surveillance systems, patient monitoring, human-computer interaction, just to name a few. Numerous techniques have been developed solve this problem 2D and 3D spaces. However imaging gained lot interest nowadays. In paper we propose novel view-independent algorithm based on fusion between global feature graph feature. We used the motion history skeleton volumes; compute for each volume action. Then, alignment performed using cylindrical coordinates-based Fourier transform form vector. A dimension reduction step subsequently applied PCA classification carried out by Mahalonobis distance, Linear Discernment analysis. The second temporal changes bounding volume, volumes are aligned divided into sub then change calculated classified Logistic Model Trees. done majority vote. proposed technique evaluated benchmark IXMAS i3DPost datasets where results compared against individually. Obtained demonstrate that improve accuracy over individual features can be recognize human actions independent view point scale.