基于复杂强度分布量测随机集模型的扩展目标跟踪方法研究

项目名称： 基于复杂强度分布量测随机集模型的扩展目标跟踪方法研究

项目编号： No.61203220

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 自动化学科

项目作者： 朱洪艳

作者单位： 西安交通大学

项目金额： 23万元

中文摘要： 随着高分辨率传感器的广泛使用，以及近距对地对海探测需求，传感器能分辨出同一目标的多个散射单元。研究针对扩展目标的建模与跟踪方法是目标跟踪领域的必然要求。受扩展目标复杂的量测产生机理、以及量测源与量测值对应关系不明确等因素影响，传统的适用于点目标的跟踪方法无法适应对扩展目标的跟踪需求。随机有限集方法的引入为这一领域的研究提供了新思路。针对复杂形状扩展目标跟踪问题，最核心和困难的地方体现在对扩展目标量测源进行合理建模、以及提出针对复杂量测源模型的跟踪方法方面；其科学本质是：如何恰当地表述扩展目标的量测产生机理，并结合时间演化模型与测量模型，求解扩展目标形状参数和质心状态的联合估计问题。本项目将提出基于复杂强度分布量测随机集的扩展目标跟踪方法。研究内容包括：复杂形状扩展目标量测源的随机有限集建模、基于累积等效量测的扩展目标形状参数估计方法、以及基于复杂强度分布量测随机集模型的质心状态估计。

中文关键词： 扩展目标跟踪；量测集划分；随机矩阵；概率假设密度滤波器；随机超曲面

英文摘要： Along with the wide use of modern high-resolution sensor and short-range surveillance need for ground and ocean targets, the sensor can resolve more than one scatter cell from the same target. It is necessary to develop efficient methods to model and track for the extended target.Traditional tracking methods for point targets fail to perform well for tracking the extended target, due to the complex measurenment generation scheme and the association between measurment sources and measurments. The RFS (Random finite set) theory provides a good tool for this research area. To track the extended target,the key problem is to establish a fit model for measurement sources of the extended target, and develop tracking methods based on the complex measurement sources model.Its scientific essence is how to describe the mesurement generation scheme and provide the estimates for both the centroid and target extent.The project will present an approach to track the extended target based on measurment RFS model with complex intensity distribution. Primary contents of the research include modeling for extended target with the complex shape, estimating for shape parameters based on equivalent measurements, and estimating for the centroid based on measurment RFS model with complex intensity distribution.

英文关键词： Extended target tracking；measurement set partitioning；random matrix；probability hypothesis density filter；random hypersurface model