Adaptive deformable model for automatic segmentation of vocal cords quivering image

A new hierarchical adaptive deformable model based on an iteration algorithm was proposed to segment the vocal cords quivering images automatically, which consists of two sub-models: primary model and high-order model. The primary model formed by two spline curves describing two vocal cords' inner boundary is defined and initialized according to the prior knowledge about the shape of vocal cords at first, then it is tuned by the high-order model which can capture the subtle edges of vocal cords by increasing the curves' order and the number of control points. The experimental results show the good performance and robustness of the proposed model since it takes full advantage from the prior knowledge and global information of vocal cords.