一种鲁棒的摄像机运动分类算法

A Robust Approach to Camera Motion Classification

摄像机运动分类是基于内容的视频分析和理解的重要问题.本文通过对运动矢量场的分析,提出了一种基于统计学习的、分层次的摄像机运动分类算法.该算法利用支持向量机(SVM)在有限样本条件下的学习能力,实现摄像机运动类型的初步分类;然后,充分考虑运动矢量场的方向和位置信息,进一步区分缩放和旋转操作,并识别摄像机平移操作的方向.算法在运动矢量的预处理过程中引入摄影规则,有效地降低了前景运动噪声的影响.

Camera motion classification is an important issue in content-based video analysis. In this paper, a robust and hierarchical camera motion classification approach based on statistical learning is proposed. As Support Vector Machines （SVM） has a very good learning capacity with limited sample set without incorporating problem domain knowledge, in the first step, SVM is employed to classify camera motion operations into two classes：translation and nontranslation operations. Then, rotation and zoom operations are distinguished using motion vectors location and direction. The direction of translation operation is also identified. In the pre-processing step, cinematic rule is utilized to filter atypical noise and foreground motion noise.