We propose a real-time 3D model-based method that continuously recognizes dimensional emotions from facial expressions in natural communications. In our method, 3D facial models are restored from 2D images, which provide crucial clues for the enhancement of robustness to overcome large changes including out-of-plane head rotations, fast head motions and partial facial occlusions. To accurately recognize the emotion, a novel random forest-based algorithm which simultaneously integrates two regressions for 3D facial tracking and continuous emotion estimation is constructed. Moreover, via the reconstructed 3D facial model, temporal information and user-independent emotion presentations are also taken into account through our image fusion process. The experimental results show that our algorithm can achieve state-of-the-art result with higher Pearson's correlation coefficient of continuous emotion recognition in real time.

The deployment of smallcell eNodeBs (SeNBs), overlaid on existing macrocell eNodeB (MeNB) is widely accepted as a key solution for improving spectral efficiency (SE). However, both SeNB and MeNB may suffer significant performance degradation due to inter/intra-tier interference. Meanwhile, the energy efficiency (EE) is another promising requirement especially when SeNB/MeNB are densely deployed. In this paper, a utility function with an α-adjustable parameter is proposed to achieve an optimal tradeoff between EE and SE. Then, an energy-aware capacity offload between the MeNB and multiple SeNBs is formulated as a Nash bargaining game, which is significantly simplified during the following analysis. To attain a win-win optimality for both the relatively involved SeNBs and MeNB, an energy-aware trigger of source-MeNB, interference-related right selected target-SeNB, and mutual interference compensation are all provided in this paper, which help to attain more dimensions of diversities and gains. Finally, simulation results show the improved performance of our proposed scheme.