A Target Tracking Model for Enterprise Production Monitoring System Based on Spatial Information and Appearance Model
Xun Li, Chuan Lin, Xinpeng Xu · Traitement du signal · 2019
Target tracking is the key function of intelligent production monitoring systems in manufacturing enterprises.In many monitoring systems, however, the fields of view (FOVs) of multiple cameras only partially overlap each other, leaving many blind zones that affect the target tracking effect.To solve the problem, this paper explores the spatial correlation probability of appearance features, and discusses about the brightness transfer function (BTF) between different FOVs.Next, the effects of appearance matching were improved, the target position was reconstructed in world coordinate system, and the spatial features of the target was extracted.On this basis, the author established a target tracking and matching appearance model with spatial information.Finally, the effectiveness of the model was fully verified through a simulation on the multi-camera production monitoring system of a factory.