学术报告

【online】Minimal Paths for Tubular Structure Tracking

发布人:发布时间: 2021-03-17

字体大小: 【小】 【中】 【大】

   Minimal Paths for Tubular Structure Tracking


:陈  副教授


   :山东省人工智能研究院


报告时间2021319日上午10:30


报告方式:腾讯会议   ID: 261 408 179


摘要The minimal geodesic technique was introduced about 20 years ago as an efficient tool for image analysis through the Eikonal PDE framework. As an important advantage, the minimal geodesic model is capable of finding a globally minimizing curve, thus can avoid unexpected local minima. Minimal geodesic paths have been used for long to carry out the tracking tubular structure tasks in a great variety of scenarios, where the features of interest, including tubular centerlines and the corresponding boundaries, can be modeled as geodesic paths. The solutions can be regarded as a way to find a curve globally minimizing the weighed curve length energy associated to tubular features.

In this talk, we first present the mathematical background for local geodesic metrics and the Eikonal PDE framework. Then we will show that various tubular features such as appearance features, anisotropic and/or asymmetric features, and prescribed tubular trajectories, can be exploited for geodesic paths extraction. Finally, we show some examples for the applications of those minimal paths in vessel tree extraction.

 

邀请人:杨孝平 老师