Local directional-structural pattern for person-independent facial expression recognition

Abstract

Existing popular descriptors for facial expression recognition often suffer from inconsistent feature description, experiencing poor accuracies. We present a new local descriptor, local directional-structural pattern (LDSP), in this work to address this issue. Unlike the existing local descriptors using only the texture or edge information to represent the local structure of a pixel, the proposed LDSP utilizes the positional relationship of the top edge responses of the target pixel to extract more detailed structural information of the local texture. We further exploit such information to characterize expression-affiliated crucial textures while discarding the random noisy patterns. Moreover, we introduce a globally adaptive thresholding strategy to exclude futile flat patterns. Hence, LDSP offers a stable description of facial expressions with the explicit representation of the expression-affiliated features along with the exclusion of random futile textures. We visualize the efficacy of the proposed method in three folds. First, the LDSP descriptor possesses a moderate code-length owing to the exclusion of the futile patterns, yielding less computation time than other edge descriptors. Second, for person-independent expression recognition in benchmark datasets, LDSP demonstrates higher accuracy than existing descriptors and other state-of-the-art methods. Third, LDSP shows better performance than other descriptors against noise and low resolution, exhibiting its robustness under such uneven conditions.

Dynamic Facial Expression Understanding Using Deep Spatiotemporal LDSP On Spark

“…However, in recent work, the top two edge responses achieve better performance [44]. We operate our proposed descriptor in three orthogonal planes (represented as LDSP-XY, LDSP-XT, and LDSP-YT) to capture both spatial and dynamic features in the RGB channels, whereas LDSP [19] extracts only the spatial information from gray images.…”

“…Then, we perform feature extraction and feature learning by employing the proposed feature descriptor and model. In our work, we introduce a novel dynamic feature descriptor, namely, the local directional structural pattern from three orthogonal planes (LDSP-TOP), which is an extension of the local directional structural pattern (LDSP) [19]. LDSP-TOP extracts the spatiotemporal information from the videos.…”

Face Recognition Based on Local Gradient Number Pattern and Fuzzy Convex-Concave Partition

“…To improve the discrimination ability, LDTP [26] integrates the local direction and the local gray-level information in its coding scheme. Recently, the local directional-structural pattern (LDSP) [32] and the local directional value (LDV) [33] have been proposed. The former uses the position and direction relationship of the edge response values to extract more detailed structural information, and the latter reduces the influence of grayscale variation and noise by combining multi-scale analysis.…”

“…In order to evaluate the performance of the proposed operator FCCP_LGNP, the LDP-based methods are chosen for comparison firstly, including LDP [23], ELDP [24], LDN [25], LDTP [26], MGP [19], LDSP [32], LDV [33] and the central pixel selection (CPS) strategy [39]. For CPS strategy, it was fused with ELDP (called CPS_ELDP in the paper).…”

“…For CPS strategy, it was fused with ELDP (called CPS_ELDP in the paper). We test the approaches on four face databases, ORL [40], CALTECH [41], GEORGIA [42], and FACE94 [44] in the experiment, and (R l , 8t) was set (1,8), (2,16), (3,24), (4,32), (5,32), (6,32) and (7,32), respectively. The Nearest Neighbor Classifier (NNC) was chosen for face recognition.…”