Research On Specific Moving Object Tracking And Real-time Distance Measurement Based On Binocular Vision

It is a thot issue that tracking and distance measurement for moving objects based on binocular stereo vision in current computer vision research.It has extensive applications in military and civilian fields.At present,the foreign measure distance products based on binocular vision of moving objects offers high accuracy and strong real-time performance,but the price is very expensive.There is still a certain gap between the domestic and overseas similar products in measurement accuracy and real-time.Firstly,in view of the low accuracy of the traditional detection algorithm,improved the background model update strategy,according the difference between the current picture and the initial background picture of each pixel.It has been proved that improved algorithm has high accuracy by experiments.Then,the tracking algorithm of kernelized correlation filters is adopted to enhance the efficiency of target tracking.However,the tracking algorithm also has some problems causing by the inability to adapt to scale change or long-term occlusion.These problems are solved by adding an online detector and improving target template update strategy.In the binocular stereo matching part,a non-local stereo matching algorithm based on the minimum spanning tree is used in this paper.The algorithm has the advantages of strong real-time and low computational complexity,owing to using bilateral filtering and the minimum spanning tree in the part of matching cost aggregation.The effectiveness of' the matching algorithm is verified by experiments in the actual environment.Finally,the experimental platform is built by using the binocular camera and the computer.The software programming of the tracking and depth extraction of the moving target is designing and achieving,and the experimental data is analyzing to verify the validity and feasibility of the algorithm.The experimental test shows that the average error of the algorithm is 5mm within 1-4 meters,and the time is about 2000ms.Real-time and accuracy meet the needs of practical application.