摘要： 在综合视觉运动分析中的两类处理方法，选取图像中的角点作为特征点，并检测和跟踪图像序列中的角点。记录检测到的角点在图像序列中的位移，在理论上证明了时变图像的光流场可以近似地用角点的位移场代替，同时给出这种替代的两个前提条件。本文用真实图像序列验证提出的算法，实验结果表明该算法取得了较好的效果。

关键词: 角点检测, 角点跟踪, 位移场, 光流场

Abstract: Motion analysis can be roughly classified as feature-based or flow-based, according to if the data they use are a set of features matches or an optical flow field. The paper integrates the two methods by using the corner points as feature points and estimating sparse optical flow field. Firstly, detecting and tracking corner points in image sequence. Secondly, the locations of tracked corner points are recorded. Finally, the projective pinhole camera model is adopted, camera motion modeled as a rigid body motion and characterized by a linear and a rotational velocity vectors. The optical flow field is estimated at sparse locations by measuring the displacement of corner points between consecutive frames. The displacement is shown to be a realistic approximation under these assumptions. Experimental results shows that the new method provides a good estimation of the optical flow field.

中图分类号: 

• TP391.41