用于彩色目标跟踪的改进粒子群优化算法

An improved PSO for color target tracking

提出一种用于彩色目标跟踪的改进粒子群优化算法(Improved Particle Swarm Optimization Algorithms,IP-SOA)。针对彩色目标,选择加权彩色直方图作为目标的特征,选用Bhattacharyya系数作为特征相似性度量,其最大值位置表示目标位置。对粒子群优化算法进行了改进,即自动调整惯性权重函数与认知学习因子,每次递推时对粒子速度、单帧位移总量加以限制,对Bhattacharyya系数优化,快速求取函数最大值位置。利用彩色序列图像进行仿真实验,结果表明,该方法能够实时跟踪飞机、车辆等目标,在目标被部分遮挡时能稳健跟踪。

An improved PSO for the tracking of color target is presented.The proposed technique employs a color-based histogram with different weight as the target feature,which is measured by Bhattacharyya coefficient.The basic PSO is improved through auto-regulative inertia weight and acceleration coefficients,restricting the motion for each frame and velocity for each iteration step,which is for fast locating the maximum of Bhattacharyya coefficient.The simulation experiments show that the method is effective and robust to the target with partial occlusion,such as plane,vehicle,etc.