Multi-feature cascade fusion related particle filtering video tracking method

• Main Authors: LIANG ZHONGHUA, YANG XIAOJUN, JIAO QINQIN, XU XIANFENG
• Format: Patent
• Language: Chinese; English
• Published: 19.04.2022
• Subjects: CALCULATING; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a multi-feature cascade fusion related particle filtering video tracking method, and aims to solve a series of challenging problems such as target deformation and shielding in a visual tracking process and the problem of weak robustness of single-feature target tracking. A correlation filter guides sampled particles to a target state distribution mode by using a correlation particle filtering algorithm framework. Meanwhile, a series hierarchical fusion mode and a two-stage particle weight updating mode are adopted. Therefore, the calculation complexity of particle filtering is reduced, and the real-time performance and the accuracy of visual tracking are improved. Under the influence of challenging factors such as shielding, illumination change and scale change, the target can still be tracked with high precision, and higher robustness is shown. 本发明公开了一种多特征级联融合的相关粒子滤波视频跟踪方法，针对视觉跟踪过程中目标形变和遮挡等一系列挑战性问题及单一特征目标跟踪鲁棒性弱的问题。利用相关粒子滤波的算法框架，相关滤波器将采样的粒子引导到目标状态分布的模式。同时采用串联分级融合方式及两级粒子权值更新方式。从而降低了粒子滤波