Facial expression recognition has recently received an increasing attention due to its great potentiality in real world applications. Conventional facial expression recognition is often conducted on the assumption that training data and testing data are obtained from the same dataset. However, in reality, the data are often collected from different devices or environments, which will severely degrade the recognition performance. To tackle this problem, in this paper, we investigate the cross-dataset facial expression recognition problem, and propose a novel dual-graph regularized transfer sparse coding method (DGTSC). Specifically, aiming to reduce the distribution divergence of different databases while preserving the geometrical structures, we construct a dual-graph, by defining the inter-domain and intra-domain similarity, to measure the distance between different databases. Moreover, we further present a dual-graph regularized discriminative transfer sparse coding method (DGDTSC), which exploits the label information, to make our model has more discriminative power. Extensive experimental results and analysis on several facial expression datasets show the feasibility and effectiveness of the proposed methods.

面部表情识别由于其在现实应用中的巨大潜力，最近受到越来越多的关注。传统的面部表情识别通常是在假设训练数据和测试数据是从同一数据集中获得的前提下进行的。但是，实际上，通常是从不同的设备或环境中收集数据，这将严重降低识别性能。为了解决这个问题，在本文中，我们研究了跨数据集的面部表情识别问题，并提出了一种新颖的双图正则化转移稀疏编码方法（DGTSC）。具体来说，为了在保留几何结构的同时减少不同数据库的分布差异，我们通过定义域间和域内相似性来构造对偶图，测量不同数据库之间的距离。此外，我们还提出了一种利用图谱信息的双图正则化鉴别转移稀疏编码方法（DGDTSC），使我们的模型具有更大的鉴别能力。大量的实验结果和对几个面部表情数据集的分析表明，该方法的可行性和有效性。