Infrared Target Tracking Algorithm Based On Mean Shift

With the increasing development of computer vision technology in today’s society, the infrared image processing technology has been used in a wide range of practical applications of navigation tracking, artificial intelligence and automatic control, infrared target problem has profound significance and has been widely used for example in the military, infrared moving target tracking has been successfully used for imaging-guided weapons, military reconnaissance and surveillance areas, in everyday life, infrared target tracking is also often used in intelligent transportation, video surveillance and other issues. This paper mainly studied the mean shift algorithm in infrared target tracking them and to study the infrared target tracking algorithm in actual use drawbacks and shortcomings that exist in the scene improved algorithm.This paper introduces the basic principles of the mean shift algorithm, the classical algorithm implementation in infrared image sequence tracking. By using classical algorithm sets of infrared target image sequence experiments, we found that the mean shift algorithm can be applied to change the size or obscured to some extent, target tracking scenarios in case the effect is not good, but has a very small mean shift algorithm computation the excellent performance in the target tracking problem can achieve a good real-time.Secondly, some of the research to know the mean shift algorithm is applied to the disadvantage of infrared image sequence, the paper using the brightness- infrared target characterization and gray space under the distance-weighted nuclear likelihood ratio histograms describe ways to improve characterization of infrared target, then the classical optimization algorithms enable it to successfully target tracking infrared target tracking problem size increases or decreases. Experiments found that the improved algorithm can be updated template, can be very accurately track changes to the size of the target, but the effect is not ideal when tracking fast moving target or undergoing complex changes in posture.In the final chapter of the article, and briefly discusses the theory and particle filter infrared target tracking applications, through two sets of experiments to prove the relationship between the beam and the computation time of the particle filter tracking anti-blocking ability and particle filter. Because of the particle filter can be a good solution to the nonlinear, non-Gaussian problem, this paper uses a blend of mean shift and particle filter method of infrared target tracking algorithm, after several different sets of experiments to know that the fusion algorithm for infrared target tracking has good accuracy and robustness.