Target Tracking System Based on Dynamic Scene

Abstract

Abstract Under the scenario of moving target tracking is very important in the fields of computer vision and research direction, its purpose is to give computer vision, double eye for simulation of the human visual information processing and information feedback process, in the process of the research can obtain the target trajectory information and provide related data for target tracking, The relevant research results of target tracking have been successfully developed in many fields such as human-computer interaction, security monitoring, intelligent monitoring and face recognition. However in the scene moving target vulnerable to disorder interference, light, shade, and scale complex situations, such as the change of the apparent dynamic change in the process of movement target, research can achieve different scenario stable tracking method has important research significance and engineering related events in the field of value, the existing popular nuclear-related filter tracking algorithm was improved, To solve the problem of occlusion and multi-scale change of tracking target, a new tracking algorithm based on multi-feature and multi-scale anti-occlusion is proposed by using kernel correlation filtering method.