Active Contour Model And Its Application On Face And Ear Contour Extraction

Active contour Model (ACM) is a new type of contour extraction method, Comparing with other traditional methods, it has remarkable extracting and tracking ability in specific area or contour . It has been widely adopted in image segmentation and contour-extraction field. For example , it is used by extracting contour in aerospace and medical images.The disadvantages of ACM are in two aspects: an initial contour should be given in order to ensure model work properly; different models should be corresponded to different issues. We proposed some improvements for these disadvantages and challenged the model on biometric feature extraction in this thesis, and present two new algorithms respective in static face shape automatic extraction under complex background and automatic auricle contour extraction under simple background.On the aspect of automatic face extraction , we first improve the model on one order continuous energy item, curve rate parameter item in the energy equation and parameter automatic adopting in iterative computing,etc. Second, we compare RGB K-L, HSV and YUV color spaces by experiments, and put forward our algorithm. The algorithm preprocesses the sample image by YUV color model, confirms the face region with morphologic features and eignface template, and finally implements automatic face contour extraction under complex background by using revised ACM model mentioned above. Experiment results show the algorithm is feasible.On automatic auricle contour extraction, we also present our algorithm. It adopts the improved single threshold otsu segmentation method to binary of the input and fits the curve to solve distinct contour in the tragus. Then we use the integrated curve as the initial input of ACM and finally get a coarse contour. Experiment results show that the algorithm is effective and accurate.