基于瑞利分布的粒子滤波跟踪算法
doi: 10.3724/SP.J.1146.2012.01225
A New Particle Filter Tracking Algorithm Based on Rayleigh Distribution
-
摘要: 粒子滤波是一种非线性滤波算法,在目标跟踪中得到了广泛的应用。基本的粒子滤波算法中粒子散布是通过高斯分布实现的,对于低速运动且变化平缓的目标可以较好地捕获,但是当目标运动状态突然改变时(如突然加速、减速或转弯等),由于高斯分布曲线的拖尾比较轻,粒子散布范围比较小,经常发生目标丢失的现象。针对该问题,该文提出使用瑞利分布来散布粒子,实现粒子的繁殖与传播,同时根据目标的运动速率对瑞利分布参数进行自适应调节,并通过二次粒子滤波方法提高跟踪精度。实验证明,基于瑞利分布的粒子传播机制可以有效提升粒子滤波的目标捕获能力。Abstract: The particle filter is a nonlinear filtering algorithm and it is widely used in target tracking. In the basic particle filter, the particles are scattered by Gaussian distributions, and the target for low-speed movement and gentle changes can be better captured. However, when the motion status of target suddenly changed, such as sudden acceleration, deceleration, or turning, etc., it often leads to the target loss. The reason is that the tailing of the Gaussian distribution curve is relatively lighter than that of the Rayleigh distribution, and then it causes the distributing range of particles is too small to capture the target. Considering this issue, the Rayleigh distribution is proposed to spread the particles, and then achieve the particles reproduction and dissemination, the Rayleigh distribution parameters is adaptively adjusted according to the rate of movement of the target, and through the secondary particle filter method to improve tracking accuracy. The experiments demonstrate that the particle propagation mechanism based on Rayleigh distribution can effectively enhance the ability of capturing target.
-
Key words:
- Target tracking /
- Particle filter /
- Particle propagation /
- Rayleigh distribution /
- Gaussian distribution
-
计量
- 文章访问数: 2650
- HTML全文浏览量: 114
- PDF下载量: 1363
- 被引次数: 0