A real-time moving target tracking algorithm based on SIFT

Abstract

Moving target tracking is an important research area in pattern recognition, image processing and computer vision. We proposed a novel moving target tracking algorithm based on SIFT in this paper. For the high running time and large computation amount of the traditional SIFT algorithm, we design a new keypoint descriptor, reduce the keypoint dimension, improved the keypoint extraction rate. It is proved by experiments that the new algorithm meet the real-time requirement and can tracking the occludent moving target well.