基于视差信息的立体图像主动闭合轮廓提取算法研究

针对立体图像中目标对象的闭合轮廓提取任务，该文提出一种基于视差信息的轮廓提取算法。该算法在传统贪婪蛇模型的基础上，利用各控制点和中心的视差关系为模型设计收缩和膨胀力，能够有效指导初始轮廓曲线向目标边缘的收敛。同时算法采用重复利用经处理的控制点作为模型输入的循环迭代方式，能够获得分布均匀且较为密集的边缘曲线。实验结果表明，该轮廓提取算法减少了传统的贪婪蛇模型算法对初始值的依赖，准确度和可靠性均得到了很大的提升。

Extraction Algorithm for Active Contour Based on Disparity Information of Stereoscopic Image

To get the closed boundary of target object of given stereoscopic image, this paper proposes an improved algorithm with disparity-guided constraint for active contour extraction based on greedy snake model. The disparity relationship between the control points and the center of the object is used to design a new external force, which guides the original contour to real boundary. In addition, this method also uses the improved model iteratively with processed control points as the original input. Experiment results prove this method reduces the reliance on initial values of input contour, obtains accurate contour made up of uniform and dense points, which is more effective than the traditional greedy snake algorithm.