Prior Shape Acquisition And In The Face Image Segmentation Applications

Digital image processing technology has been developed widely in recentyears,it also has been applied to every aspect of social development and humanlife rapidly.Image segmentation plays a vital role in the process from imageprocessing to image analysis and completing the process of image. Based onlevel set,geometric active contour model can handle the change of topologicalstructure naturally,it can obtained the target contour segmentation of the idealsegmentation results. The level set image segmentation algorithm based on priorshape constraint is an important research direction in recent years. A priori shapeacquisition is an important approach to extract target in complex image (coveredimage, data loss,etc.)This paper illustrates the level set of the related theory、classic MumfordShah model and Chan-Vese model.Based on losing data image and obscuredimage,classic model don’t have a good segmentation effect，we have did thefollowing two points：(1)A new math segment model is constructed，whichJoined local invariant to X、Ydirection tensile and shear properties based on CVmodel basis；(2)A new construction method of shape energy term that makesevolving surface stable is presented，which considers the global and localchange.Image segmentation and transcendental templates under the condition oflittle change, the model can have a good segmentation effect. In view of the image face expression have a larger change, the above model that is have asingle shape template segmentation result is not ideal, therefore, this articlebased on PCA established a parameter adaptive KPCA (Kernel PrincipalComponent Analysis) of prior shape constraint level set mathematical model.Before segmentation using KPCA extracting nonlinear characteristics in trainingof library image, when iterative segmentation can be trained to focus on anyimage evolution rather than a single template. Finally according to parameterselection time-consuming problem, design the related algorithm,which can beadaptive to choose the parameters and improve the rate.