Lucas-Kanader-Tomasi Feature Tracker，由运动恢复结构的问题，目的是从一组由摄像机移动获得的图像来恢复一个场景的立体结构。对于这个问题，C.Tomasi和T.Kanade[1]于1992年提出了一种Factorization的方法，可以有效地避免噪声的干扰，同时不会受限于特定的运动模型，例如单纯的平移或是旋转。-Lucas-Kanader-Tomasi Feature Tracker, by the movement to restore the structure of the problem, aimed at a group of mobile access by the camera image to restore the three-dimensional structure of a scene. For this problem, C. Tomasi and T. Kanade [1] in 1992 proposed a factorization method which can effectively avoid noise interference, while not restricted to a particular motion model, such as pure translation or rotation .