Target Tracking Algorithm Based On Local Invariant Features

With the rapid development of computer vision and image processing technology,target tracking has been a hotspot and difficult in the field of application research in recent years.Target tracking is of great significance in the fields of medicine,industry,navigation,information processing and so on.It is always the highest goal of video tracking to study a better real-time,strong in robustness and higher in versatile target tracking algorithm.In order to improve the robustness of the target tracking algorithm,the local invariant feature descriptor of the target can be used to ensure the accuracy of the tracking of the moving target in the case of scale change,illumination change and local occlusion.Scale invariant feature change is the most representative local descriptor in local invariant features.The SIFT feature is robust to changes in scale,scaling and rotation,and has a strong invariance for viewing angle changes,illumination changes and occlusion.Therefore,it has always been a hot topic in video tracking.This subject mainly studies the target tracking algorithm based on SIFT feature.It is characterized by the SIFT feature because the SIFT feature is very unique and can be quickly and accurately matched in the massive feature data;and it has a stable invariance when the target changes.Based on the characteristics of SIFT feature,the algorithm also has good stability and strong robustness.However,when the algorithm deals with large images or images containing rich information,the algorithm is not real-time and can be wasted resources in practical applications.Aiming at these shortcomings,this paper proposes a target tracking algorithm based on Mean-Shift based on SIFT feature point extraction.The algorithm uses the Mean-Shift algorithm to predict the position of the moving target.By predicting the initial position of the moving target,the local SIFT feature detection domain is established to realize the rapid extraction and matching of the SIFT feature and further locate the position of the moving target.This motion prediction-based tracking algorithm narrows the search range of moving objects and greatly improves the speed of motion.In this paper,we use MATLAB to simulate the improved algorithm and the traditional Mean-Shift algorithm,and analyze and compare the experimental results.The improved target tracking algorithm can accurately track the position of the moving target in the tracking process.Compared with the unmodified algorithm,it can be used to save a lot of resources and resources in the real tracking time.