少量交互的多视角图像目标分割算法.pdf

第29 卷 第 1 期 计算机辅助设计与图形学学报 Vol. 29 No.1 2017 年 1 月 Journal of Computer-Aided Design Computer Graphics Jan. 2017 少量交互的多视角图像目标分割算法 缪 君, 储 珺, 张桂梅 (南昌航空大学江西省图像处理与模式识别重点实验室 南昌 330063) (miaojun@) : 目前, 单一图像在人工交互的条件下已能实现准确分割, 但如果图像数量大, 则需要大量劳动量. 而自动分 割难以在多图像间构建统一的目标模型. 为此, 提出一种多视角图像的少量交互分割算法, 即仅对其中一幅进行交 互, 对其他图像进行前景目标一致的自动分割. 首先通过构建特征点轨迹图模型, 建立其他图像特征点与交互图像 特征点之间的连接; 然后利用最短路径算法进行特征点的前景/背景分类, 获得多图像间的局部对应; 最后将已分类 特征点作为形状先验, 使用测地星凸分割算法实现多视角图像的一致分割. 命中率和误报率的比较实验结果表明, 文中的算法具有较好的一致分割能力, 而且这种局部对应的方式还可以实现多目标的分割. ：多视角图像; 目标分割; 交互; 特征点 ：TP391.41 Object Segmentation from Multi-Views Images with a Few Interactions Miao Jun, Chu Jun, and Zhang Guimei 1) (Key Laboratory of Image Processing and Pattern Recognition, Nanchang Hangkong University, Nanchang 330063) Abstract: With the help of human-computer interaction (HCI), a monocular image can be segmented accu- rately. However, manual workload of HCI is remarkably heavy if a number of images are pending processing. Furthermore, it is difficult to construct a unified model that is universally applicable to different images. We propose an accurate segmentation algorithm for multi-view serial images with less HCI work. HCI is per- formed only on one serial image, and then the objects in the foreground of the other images can be seg- mented automatically. Based on the homography of the feature points built with a graphic model, the feature points in the HCI image and their corresponding points in other serial images are accordingly linked. The shortest path algorithm classifies the feature points into foreground or background, and searches the corre- spondence between the local fea