Tracking Features with Large Motion

This paper addresses feature tracking when frame-toframe motion is too large that the popular pyramidal Kanade-Lucas-Tomasi (KLT) feature tracker does not work. To solve this problem, we estimate the motion at the deepest pyramid level by matching the horizontal (and vertical) characteristic curves of the consecutive images. To compute the motion estimates efficiently and effectively, we use dynamic programming to minimize the cost function. These motion estimates will serve as the coarse motion at the deepest pyramid level which makes the residual motion small enough such that the feature tracker can work well. Experiments show that our method can make the feature tracker suitable for features with large motion.