Exploring visual features through Gabor representations for facial expression detection

Gabor representations have been widely used in facial analysis (face recognition, face detection and facial expression detection) due to their biological relevance and computational properties. Two popular Gabor representations used in literature are: 1) Log-Gabor and 2) Gabor energy filters. Even though these representations are somewhat similar, they also have distinct differences as the Log-Gabor filters mimic the simple cells in the visual cortex while the Gabor energy filters emulate the complex cells, which causes subtle differences in the responses. In this paper, we analyze the difference between these two Gabor representations and quantify these differences on the task of facial action unit (AU) detection. In our experiments conducted on the Cohn-Kanade dataset, we report an average area underneath the ROC curve (A`) of 92.60% across 17 AUs for the Gabor energy filters, while the Log-Gabor representation achieved an average A` of 96.11%. This result suggests that small spatial differences that the Log-Gabor filters pick up on are more useful for AU detection than the differences in contours and edges that the Gabor energy filters extract.

Large margin vector quantization

In this paper we describe the Large Margin Vector Quantization algorithm (LMVQ), which uses gradient ascent to maximise the margin of a radial basis function classifier. We present a derivation of the algorithm, which proceeds from an estimate of the class-conditional probability densities. We show that the key behaviour of Kohonen's well-known LVQ2 and LVQ3 algorithms emerge as natural consequences of our formulation. We compare the performance of LMVQ with that of Kohonen's LVQ algorithms on an artificial classification problem and several well known benchmark classification tasks. We find that the classifiers produced by LMVQ attain a level of accuracy that compares well with those obtained via LVQ1, LVQ2 and LVQ3, with reduced storage complexity. We indicate future directions of enquiry based on the large margin approach to Learning Vector Quantization.

Expression of chondromodulin-1 in the temporomandibular joint condylar cartilage and disc

BACKGROUND: The temporomandibular joint (TMJ) cartilage consists of condylar cartilage and disc and undergoes continuous remodeling throughout post-natal life. To maintain the integrity of the TMJ cartilage, anti-angiogenic factors play an important role during the remodeling process. In this study, we investigated the expression of the anti-angiogenic factor, chondromodulin- 1 (ChM-1), in TMJ cartilage and evaluate its potential role in TMJ remodeling. METHODS: Eight TMJ specimens were collected from six 4-month-old Japanese white rabbits. Safranin-O staining was performed to determine proteoglycan content. ChM-1 expression in TMJ condylar cartilage and disc was determined by immunohistochemistry. Three human perforated disc tissue samples were collected for investigation of ChM-1 and vascular endothelial growth factor (VEGF) distribution in perforated TMJ disc. RESULTS: Safranin-O stained weakly in TMJ compared with tibial articular and epiphyseal cartilage. In TMJ, ChM-1 was expressed in the proliferative and hypertrophic zone of condylar cartilage and chondrocyte-like cells in the disc. No expression of ChM-1 was observed in osteoblasts and subchondral bone. ChM-1 and VEGF were both similarly expressed in perforated disc tissues. CONCLUSIONS: ChM-1 may play a role in the regulation of TMJ remodeling by preventing blood vessel invasion of the cartilage, thereby maintaining condylar cartilage and disc integrity.

Capital music : personal expression with a public display of song choice

Using information and communication technology devices in public urban places can help to create a personalised space. Looking at a mobile phone screen or listening to music on an MP3 player is a common practice avoiding direct contact with others e.g. whilst using public transport. However, such devices can also be utilised to explore how to build new meaningful connections with the urban space and the collocated people within. We present findings of work-in-progress on Capital Music, a mobile application enabling urban dwellers to listen to music songs as usual, but also allowing them to announce song titles and discover songs currently being listened to by other people in the vicinity. We study the ways that this tool can change or even enhance people’s experience of public urban spaces. Our first user study also found changes in choosing different songs. Anonymous social interactions based on users’ music selection are implemented in the first iteration of the prototype that we studied.