Inexact Graph Matching for Facial Feature Segmentation and Recognition in Video Sequences: Results on Face Tracking
This paper presents a method for the segmentation and recognition of facial features and face tracking in digital video sequences based on inexact graph matching. It extends a previous approach proposed for static images to video sequences by incorporating the temporal aspect that is inherent to such sequences. Facial features are represented by attributed relational graphs, in which vertices correspond to different feature regions and edges to relations between them. A reference model is used and the search for an optimal homomorphism between its corresponding graph and that of the current frame leads to the recognition.
KeywordsVideo Sequence Facial Feature Model Image Graph Match Relational Graph
- 1.Colliot, O., Tuzikov, A.V., Cesar Jr., R.M., Bloch, I.: Approximate reflectional symmetries of fuzzy objects with an application in model-based object recognition. In: Fuzzy Sets and Systems (2003) (in press)Google Scholar
- 3.Cesar Jr., R.M., Bengoetxea, E., Bloch, I.: Inexact graph matching using stochastic optimization techniques for facial feature recognition. In: 16th International Conference on Pattern Recognition, August 2002, vol. 2, pp. 465–468 (2002)Google Scholar
- 4.Cesar Jr., R.M., Bloch, I.: First results on facial feature segmentation and recognition using graph homomorphisms. In: Proc. VI Simpósio Ibero-Americano de Reconhecimento de Padrões, Florianópolis, Brazil, pp. 95–99 (2001)Google Scholar
- 5.Cesar Jr., R.M., da, L., Costa, F.: Shape Analysis and Classification – Theory and Practice, 1st edn. CRC Press, Boca Raton (2001)Google Scholar
- 6.Kruger, V., Sommer, G.: Affine real-time face tracking using a wavelet network. In: Proc. of ICVV 1999 Workshop Recognition, Analysis, and Tracking of Faces and Gestures in Real-Time Systems, Corfu, Greece, pp. 141–148 (1999)Google Scholar