Active shape models their training and application

Abstract

Nowadays, image segmentation is becoming important in image processing. Several segmentation methods have been proposed in which the techniques that are trained on examples are increasingly popular in image analysis. The model-based approach is not new in segmentation. However, the techniques that model both the shape and the gray level appearance of the object, such as Active Shape Models (ASM), Active Appearance Models (AAM), cannot be ignored. In this thesis, we have concentrated on the ASM whose algorithms and calculations have been researched. Moreover, we also provide the ASM Toolkit for demonstrating ASM’s procedures and applying it to find the boundary of an object in images. This Toolkit is designed in order to solve some tasks, including: building the models of an object that can vary,

viewing built models in which displaying the modes of variation of the shape and training information, searching correctly the target object’s boundary within a new image. With the experience using the Toolkit, users can develop their own ASM applications.