基于Snake模型的图像目标轮廓自动跟踪方法.pdfVIP

第33 卷第2期 东南大学学报( 自然 科学版) Vol33 No2 2003 年3 月 JOURNAL OF SOUTHEAST UNIVERSITY (Natural Science Edition) ar. 2003 Snake 1 2 1 1 王立功 于甬华 姜晓彤 罗立民 1 ( , 210096) 2 ( , 250117) : 针对实验数据像素灰度的分布特点, 出了 一种对目标轮廓线进行有效和可靠的搜索和 跟踪策略. 由于数据中病变组织与其邻近组织像素灰度差别相对明显,首先通过采用 一种改进的 轮廓自动跟踪方法对目标轮廓进行跟踪, 将得到的轮廓线经采样得到其离散控制点作为Snake 轮廓搜索和跟踪算法的输入,既克服了Snake 方法对初始轮廓线控制点分布的局限性, 又避免了 采用单一轮廓跟踪方法跟踪目标轮廓线的不确定性, 高了分割病变组织的速度和准确性,此方 法具有较高实用意义. : Snake 模型; 轮廓; 目标跟踪; 方向编码; 交互分割 : TP391 : A : 1001- 0505(2003 Snake model based automatic tracking method for image object contour 1 2 1 1 Wang Ligong YuYonghua Jiang Xiaotong Luo Limin 1 ( Department of Biological Science and edical Engineering, Southeast University, Nanjing 210096, China) 2 ( Department of Radiotherapy, Shandong ProvinceTumorHospital, Jinan 250117, China) Abstract : Aiming at thedistribution characteristicof pixel gray level intested data, this paper proposes a valid and reliablesearching andtracking policyfor theobject contour. Becauseof therelatively distinct con trast of gray level between unhealthy tissues and neighboring ones in data, an improved method is used to trackthe object contour firstly, thenthecontour is sampledtoget the discretecontrolling points of it and in put them intotheSnakecontourtracking algorithmsubroutine, thus avoidingthe limitation ofthe initial con tour controlling points of theSnakemethod and increasing thespeed and accuracyofthe segmentation for the unhealthy tissues. The propose