Moving Target Stable Tracking Based On Image

Motion target tracking based on images is an important research content of computer vision,which is widely used in video monitoring,pattern recognition,artificial intelligence and other fields.Although many scholars of domestic and foreign have studied target tracking technology,most target tracking algorithms are still difficult to achieve the desired effect in terms of stability,which due to the interference factors in video,such as light change,dimension change,background interference,obstacle shielding and so on.From the above,this thesis studies the above four kinds of interference factors and improves the stability of single target movement.The contributions of this thesis are concluded as follows:Firstly,this thesis analyzes the commonly used target detection methods and clarifies the judgment requirements of tracking targets based on the background of the project.Then,this thesis selects continuous frame difference method for sampling,which detects moving objects within range in the basis of manual delineation by connected area search algorithm.After that,this thesis deals with the results of the connected area search.Or rather,the algorithm searches the result of the connected area,selects the appropriate part and makes the minimum connected rectangle box of the connected area.In addition,we get the moving target area and save the parameters.Secondly,this thesis combines color,edge and texture features for iteration on the basis of feature iterative algorithm.Finally,this thesis improves the feature update of the local area,which reduces the computation and ensures the real-time performance of the algorithm.This thesis mainly discusses the stability of the target in motion.In order to track a target under the specific environment,a solution with theoretical guarantee is proposed and analyzed via simulations,with the aim to reduce tracking error and enhance tracking stability.Experimental results show that the improved algorithm has good accuracy and stability.