Title :
Enhanced snake model and modified H∞ for lips contour detection and tracking
Author :
Chin, Siew Wen ; Seng, Kah Phooi ; Ang, Li-Minn
Author_Institution :
Sch. of Electr. & Electron. Eng., Univ. of Nottingham, Semenyih, Malaysia
Abstract :
An automatic lips contour detection and tracking using an enhanced snake model and modified H∞ is presented in this paper. For the conventional snakes, the resultant contour would be converged wrongly if the initial snake point is situated far from the lip edges, or with the existence of the tongue and teeth. Consequently, a new external force based on watershed approach is proposed and absorbed into the conventional snake model to deal with the presence of tongue and teeth. Furthermore, a modified H∞, tracking system is proposed to keep track on the succeeding lips feature points as the initialization of the snake model to detect the lips contour. The modification of the conventional H∞ is based on the LST to assure that the proposed tracking approach is always stable and as the time approaches to infinity, the tracking error would be converged to zero. With both the modified H∞ that offer a better tracking capability compared to the conventional H∞ and the enhanced snakes model with the watershed based external force, the proposed system provide a more precise snakes convergence to the final desired lips contour and the overall simulation results have shown an appreciable performance.
Keywords :
edge detection; face recognition; image sequences; tracking; transforms; Lyapunov stability theory; active contour model; automatic lips contour detection; automatic lips contour tracking; enhanced snake model; image sequences; modified H∞ tracking system; watershed transformation algorithm; Feature extraction; Filtering; Image segmentation; Lips; Mathematical model; Pixel; Teeth; Active Contour Model; H∞ filter; Lips Contour Detection; Lips Tracking; Lyapunov Stability Theory; Watershed;
Conference_Titel :
Computer Applications and Industrial Electronics (ICCAIE), 2010 International Conference on
Conference_Location :
Kuala Lumpur
Print_ISBN :
978-1-4244-9054-7
DOI :
10.1109/ICCAIE.2010.5735017