Distributed Soccer Training Smart Sensors for Multitarget Localization and Tracking

This paper presents an in-depth study and analysis of the localization tracking multiple targets in soccer training using a distributed intelligent sensor approach. An event-triggered mechanism is used to drive acoustic array sensors network, which solves problem increased communication load caused by frequent microphones effectively reduces between as well energy consumption network. By designing suitable state estimation equation for source target fully utilizing measurement information its nodes neighboring nodes, next moment can be accurately predicted. A correlation filtering algorithm based on multiscale spatial co-localization proposed. In proposed algorithm, tracker contains total several subfilters with different sampling ranges. Then, this also proposes collaborative discrimination method judge response image samples each filter jointly localize online. Based this, further explores potential algorithms complex environments robust that fuses views. The cross-view geometric similarity measure multiframe pose proposed, matching effect better than single-frame similarity; solve player appearance interference, graph model-based learning players view depth features node attributes, connections edges construct graph. obtained convolutional neural network calculated simple cosine distance.