标题：A variational image segmentation model using RGB-D data
作者：Wang, Jing ;Zheng, Yongguo ;Pan, Zhenkuan ;Zhang, Weizhong ;Liu, Cunliang
作者机构：[Wang, Jing ;Pan, Zhenkuan ;Zhang, Weizhong ;Liu, Cunliang ] College of Information Engineering, Qingdao University, Qingdao, China;[Wang, Jing ;Zheng 更多
来源：Journal of Computational Information Systems
摘要：Accurate extraction of some object or a person's hand from RGB-D image is a key step for many computer vision applications, such as object detection and hand detection. In this paper, Potts model which is a general variational model for gray image segmentation is extended to the one for RGB-D image segmentation. Potts model is reformulated by updating the data item which combines depth data and probabilistic image information converted from RGB image. Algorithms based on Split Bregman method and the improved dual method are designed to speed up the computational process. Some numerical experiments show that the proposed model and algorithms can segment RGB-D image effectively and produce an accurate silhouette of the object. © 2015 by Binary Information Press.