摘要
针对鱼目标跟踪,提出了一种基于粒子滤波与仿射变换的算法。该算法引入仿射变换来建立鱼姿态变化的形状空间,同时用粒子滤波在鱼姿态变化的形状空间内进行搜索,求解鱼姿态变化的参教,然后对鱼目标进行跟踪。运用鱼的运动模型进行了实验,结果显示所提算法的跟踪轮廓在正确跟踪目标的基础上,能够对目标的形状进行更加形象的描述,较好地解决了目标在颜色上出现干扰的问题,从而有效地解决了鱼在水中游动时产生的平面旋转及短暂遮挡的问题。
A new fish tracking algorithm has been developed based on affine transform and particle filtering. Affine transformation is intro- duced to build the fish shape space,and particle-filtering is used to search in the space in order to solve the parameters of fish shape change, and finally to track targeted fish. Experiments were conducted with the guidance of fish kinetics, and the results show that the pro- posed algorithm can track the targeted fish and better describe the shape of fish, and as a result, can solve the problems of in-plane rota- tion and short-time hiding.
出处
《计算机技术与发展》
2014年第4期135-138,共4页
Computer Technology and Development
基金
国家自然科学基金资助项目(61175035)
关键词
鱼目标
仿射变换
粒子滤波
跟踪算法
fish target
affine transform
particle filtering
tracking algorithm
作者简介
侯崇明（1963-），男，硕士，研究方向为人工智能应用和图像跟踪。
