This paper presents a new approach based on extension deformation for extended object tracking (EOT). In this approach, the extension of an object is assumed to be deformed from a reference extension by moving some control points in the latter to those in the former. That is, the properties of an extension can be fully captured by the control points, given the reference extension. Thus, modeling and estimation of the extension can be reduced to those of the control points. In this way, not only various extended objects can be described conveniently, but also the extension-evolution model can be simplified. Moreover, the measurement model derived from the extended object model is linear, which largely simplifies estimation. In addition, an online adaptation method is proposed to incorporate online information into the reference extension, so the proposed approach can describe and estimate the extension accurately. Simulation results of scenarios for maneuvering EOT are given to illustrate the effectiveness of the proposed approach.

中文翻译：

---

扩展对象跟踪的扩展变形方法

本文提出了一种基于扩展变形的扩展目标跟踪（EOT）新方法。在这种方法中，通过将参考扩展中的一些控制点移动到前者中的控制点，假设对象的扩展从参考扩展变形。也就是说，在给定参考扩展的情况下，扩展的属性可以完全由控制点捕获。因此，扩展的建模和估计可以减少到控制点的建模和估计。这样不仅可以方便地描述各种扩展对象，而且可以简化扩展-演化模型。此外，从扩展对象模型导出的测量模型是线性的，这在很大程度上简化了估计。此外，还提出了一种在线适配方法，将在线信息纳入参考扩展中，因此所提出的方法可以准确地描述和估计扩展。给出了操纵 EOT 场景的仿真结果，以说明所提出方法的有效性。