Research Article

Context-Aware AAL Services through a 3D Sensor-Based Platform

Table 1

Classification rates (%) of postures at varying viewpoint angles and distances, both with and without semiocclusions.

FeatureLevel 1Level 2Level 3Level 4
DistanceLowMedHighLowMedHighLowMedHighLowMedHigh

Volumetric descriptor

View angle (°)0999694 9897969594 90 747373Without semi-occlusions
45999795979694 93 93 89737172
90989793 989793 94 93 88757470
1359795969794 9592 92 91 777572
1809996959695969592 88747474
094 9592 94 92 93 91 8887676667With semi-occlusions
4594 93 93 94 94 92 93 8786666464
909594 94 91 90 91 898686676565
13593 90 92 93 94 92 848585656767
18094 92 91 92 91 93 90 8584686466

Topological descriptor

View angle (°)0999594 989693 9793 879891 86Without semi-occlusions
45989593 9794 91 9692 86998587
9094 91 90 93 92 8994 877993 8479
1359897959693 92 9790 86958783
180999894 9594 94 989588978986
094 898193 868292 877791 8780With semi-occlusions
4591 858490 857994 847592 8379
90877869887573867968867875
13592 8675878375897671888178
18090 847791 807893 827890 8281