摘要
近年来,水上交通运输的蓬勃发展对高性能的船舶视觉跟踪方法提出越来越高的要求。为实现不同环境下船舶的有效跟踪,采用颜色特征和纹理特征进行船舶特征提取。针对船舶运动的随机性,本文在布朗运动的基础上构建船舶运动模型,通过颜色特征和纹理特征构建船舶观测模型。最终,本文成功实现了一种交互多模粒子滤波多特征自适应融合的船舶视觉跟踪方法,满足水上交通管理的应用需求。
In recent years,the rapid development of water transportation has put forward more and more requirements for the high performance ship vision tracking method.In order to realize the effective tracking of ships in different environments,the features of color and texture are used to extract the ship features.In view of the randomness of the ship movement,the ship motion model is constructed by Brown movement,and the ship observation model is constructed by the color and texture features.Finally,we successfully implemented an interactive multi-mode particle filter and multi feature adaptive fusion of ship visual tracking,based on the improved particle filter algorithm.
出处
《舰船科学技术》
北大核心
2018年第2X期133-135,共3页
Ship Science and Technology
基金
广州市高校创新创业教育项目(2017201201)；广州市高校第九批教育教学改革项目(2017F06)
关键词
特征
自适应
模型
粒子滤波
feature
adaptive
model
particle filter
