Research And Design Of Multi-target Tracking System Based On Mobile Platform

In recent years,with the continuous development of high-tech technologies such as artificial intelligence,smart chips,remote control and machine vision,intelligent products based on mobile robots have penetrated into various fields such as industrial production,daily life,and smart cities.Among them,the mobile platform-based target tracking system shows great application value in assisting military strikes,regional security,and dangerous operations.Therefore,this paper focuses on the research and design of multi-target tracking systems based on mobile platforms,and discusses four levels of problems:considering the establishment of target models under dynamic backgrounds based on mobile platform work,and considering multi-targets under dynamic backgrounds.Tracking algorithm design issues,considering the formulation of mobile platform design solutions that meet actual needs,and considering the construction and functional testing of multi-target tracking system platforms based on mobile platforms.(1)Aiming at the problem of establishing the target model in the dynamic background based on mobile platform,the target's static appearance model and dynamic motion model is given.Specifically: from the perspective of three characteristics of color,texture and corner points,a variety of methods are used to establish the appearance model of the target.A three-dimensional(3D)scene space model is established according to ground plane constraints,thus the target's position and velocity are used to establish the motion model of the target,and the model's correction and update method is given according to the movement characteristics of the mobile platform.(2)Aiming at the problem of multi-target tracking under dynamic background,the design scheme of improved POM tracking algorithm and improved KCF tracking algorithm is given.Based on the established 3D dynamic scene space model and target model,the improved POM multi-target tracking algorithm which combined with multi-feature fusion method is suitable for monocular dynamic background.The improved KCF multi-target tracking algorithm with adaptive target scale which combined with multi-core tracking method is designed.The effectiveness of the designed tracking algorithm is verified under the public data set.(3)Aiming at the need of the mobile platform formulate the corresponding mobile platform design plan,which include both hardware design plan and software design plan.STM32F103RCT6 is selected as the main control chip.Based on the functional requirements of the mobile platform,the design schemes of 4G communication,drive control,speed detection and IMU inertial detection are provided.The main program design scheme of target tracking is designed,the PID algorithm for controlling the movement of the mobile platform is given,and the PID control algorithm is experimentally verified.(4)Aiming at the platform construction and functional testing of the multi-target tracking system based on the mobile platform,a multi-target tracking system experimental platform based on the dynamic background under the C/S architecture,which include both the mobile platform and the upper PC.After setting up the experimental test environment and system operating parameters,the functions of signal transmission,target detection and multi-target tracking are tested to verify the actual application effect of the designed multi-target tracking system based on mobile platform.