Research And Application On Moving Objects Tracking Method Based On Binocular Vision

Computer vision is an important branch of computer science and artificial intelligence,important not only as an engineering tool, but also as a scientific discipline. It refers to theprocess by which images are acquired via sensors, then analyzed and processed to extract3-dimensional information. The ultimate goal of computer vision is to supplant the humanbrain as a way to model and understand reality. Binocular stereo vision technology, a sub-fieldof computer vision, is a particularly important field in modern industrial, medical, militaryand space technology fields, due to its enhanced ability to extract deep information in3-dimensional scenes. Moving target tracking technology, which utilizes binocular stereovision technology, is a comprehensive procedure that encompasses target detection, extraction,tracking, analysis and behavior comprehension.This article discusses several common methods of tracking moving targets, and explainsthe basic principles and implementation details of these methods. In particular, it proposes anobject detection method combining binocular disparity maps and multiplied framedifferences.Secondly, this method is better equipped to comprehend moving targets in3-dimensionalscenes due to the depth information provided by binocular disparity maps. Analysis showsthat this detection algorithm is able to accurately locate and identify moving objects, andfurther extract borders of said objects from sequential images.Furthermore, this algorithm introduces iteration theory and similarity measure functionsfrom mean shift theory into its particle filter algorithm, which resolves the problem of particledegradation, reduces the number of particles used and improves the algorithm’s ability totrack in real time.Lastly, this algorithm is used to track moving persons, implementing both single-andmultiple-target tracking. Analysis shows that choosing the correct tracking method isparamount in reducing tracking and computational complexity. The method is shown to behighly effective and practical in tracking objects.