Since image background is normally composed of congenial regions, it can be represented by a feature dictionary via sparse representation. Based on this theory, the authors propose a novel bottom‐up saliency detection method that unites the syncretic merits of sparse representation and multi‐hierarchical layers. In contrast to most pre‐existing sparse‐based approaches that only highlight the boundaries of a target, the proposed method highlights the entire object even if it is large. Given a source image, a multi‐scale background dictionary is structured with the features form different layers. Each region of the image is then reconstructed by the dictionary to compute its reconstruction error as a saliency score. Although a reconstruction map can be generated by the saliency scores, it is not good enough to be the final result because of low resolution and high error detection rates. Therefore, in middle cue, they propose a multi‐scale contour zooming approach to address the error detection across the hierarchical layers. To improve the resolution of the final detection, a pixel‐level rectification based on the Bayesian observation likelihood is calculated as the bottom cue. Combining sparse representation and multi‐scale correction, the precision of the final saliency map is significantly improved for the detection results.

Recently, researchers have proposed different feature descriptors to achieve robust performance for facial expression recognition (FER). However, finding a discriminative feature descriptor remains one of the critical tasks. In this paper, we propose a discriminative feature learning scheme to improve the representation power of expressions. First, we obtain a discriminative feature matrix (DFM) based pixel difference representation. Subsequently, all DFMs corresponding to the training samples are used to construct a discriminative feature dictionary (DFD). Next, DFD is projected on a vertical two‐dimensional linear discriminant analysis in direction (V‐2DLDA) space to compute between and within‐class scatter because V‐2DLDA works well with the DFD in matrix representation and achieves good efficiency. Finally, nearest neighbor (NN) classifier is used to determine the labels of the query samples. DFD represents the local feature changes that are robust to the expression, illumination et al. Besides, we exploit V‐2DLDA to find an optimal projection matrix since it not only protects the discriminative features but reduces the dimensions. The proposed method achieves satisfying recognition results, reaching accuracy rates as high as 91.87% on CK+ database, 82.24% on KDEF database, and 78.94% on CMU Multi‐PIE database in the LOSO scenario, which perform better than other comparison methods.