免疫粒子滤波在声呐图像目标跟踪中的应用

基于前视声呐图像序列,研究并实现了经免疫算法优化的粒子滤波水下目标跟踪。声呐图像分割成二值图后,提取目标的区域形状特征以构建观测模型,设计目标模板自适应更新方法;将免疫算法的克隆与变异思想引入到粒子滤波中以解决粒子退化问题。对两组水下运动物体的跟踪实验表明,即使目标存在一定形变与干扰,文中的免疫粒子滤波算法仍能以较高的精度跟踪到目标真实运动轨迹;相比于传统粒子滤波算法,稳定性也更强。

Application of immune particle filter in target tracking with sonar image

Based on the forward-looking sonar image sequences, the particle filtering (PF) optimized by the immune algorithm for underwater target tracking is studied and implemented. After the sonar image is segmented into binary image, the area shape features of the target are extracted to construct the observation model, and the adaptive updating method of the target template is designed. The clone and mutation ideas of the immune algorithm are introduced into particle filtering to solve the problem of particle degradation. The tracking experiments for two groups of underwater moving objects show that even if the target has certain deformation and interference, the immune particle filtering algorithm in this paper can still track the real trajectory with high precision and compared to the traditional particle filtering algorithm, the stability is also stronger.