The Study of Video Frame Tracking Based on the New Scale Invariant Algorithm and the Feature Classification Tree

For the difficult question of how to track moving object accurately during the augment reality without sign identification process. This paper suggests a comprehensive video frame tracking theory model. Using the new scale invariant algorithm to optimize the process of the feature extraction and the forming characteristics description. Meanwhile, Using the Feature classification tree algorithm to transform the image feature point identification matching problem into the characteristic pattern recognition classification problem¡£Finally, apply SIFT-RT model to the process of video tracking and get a good experimental results. The model owns the advantages such as the small time complexity, the precise characteristics tracking. Experiments demonstrate that SIFT-RT model can satisfy the requirements of special tracking of the augment reality without the sign.