Design And Implementation Of Aircraft Tracking System Based On Raspberry Pi

Vision is the main way for humans to obtain target information.Computer vision simulates human vision to obtain,recognize,and understand objective information.The research of target tracking technology occupies an important position in the field of computer vision,and it has a wide range of applications in social security,military reconnaissance,and autonomous driving.Aiming at the problem that the target tracking algorithm based on Kernel Correlation Filter（KCF）has an unsatisfactory tracking effect when there is interference in the background and the target is occluded,this paper uses a combination of KNN moving target detection technology and a mean shift algorithm（Mean-shift）.Algorithm,the improved KCF algorithm of occlusion detection and relocation algorithm.This article compares the commonly used motion detection methods.Among them,the background difference method is very unstable,and the Gaussian mixture model is also affected by shadows.Therefore,combined with the motion detection algorithm based on machine learning KNN background segmentation,adaptive learning according to changes in the background is effective.Automatic segmentation of moving targets.Compared with other target tracking algorithms,the KCF algorithm is currently the most widely used and stable algorithm in the target tracking field.Therefore,it is combined with the mean shift algorithm.When the response value of the KCF algorithm decreases significantly,use The mean shift algorithm further tracks the target,thereby further determining the position of the target,and uses the mean shift algorithm to correct the target position and predict the target position.Combining target occlusion detection and relocation algorithms,determine whether the tracking target is occluded,and relocate the target if it is occluded.Build an aircraft tracking system based on Raspberry Pi 4B,and experiment with the improved algorithm on the embedded platform through the collected video.Experiments show that the proposed method can accurately track the target even when the target tracking background has serious interference and the target is severely occluded,and it has a significant improvement in robustness and tracking stability.