一种基于机器学习的视频镜头边界检测方法

A Novel Approach to Video Transition Detection Based on Machine Learning

镜头边界检测是基于内容视频检索中的第一步,在视频分析中扮演着重要角色。在此基于统一的机器学习框架,提出一种新颖的模式分类方法来解决新闻和广告视频中镜头检测问题。该方法利用支持向量机将镜头分为无场景变化、切变以及大场景变化;在大场景变化中,进行快速运动和渐变的分类。同时研究了以往同类工作中所忽视的不平衡样本分类问题。实验结果表明该方法能有效检测出新闻和广告视频中的镜头转换。

Video transitio the target application of framework and proposes n detection plays an commercial detection importan in news t role in many tasks of video analysis. Aiming at video, this paper tackles the problem in a unified an alternative classification strategy. This method is made up of two phases. In the first stage, SVM is employed to classify the transitions into three classes, non-transition, cut, and big-transition. In the second stage, the focus is on the discrimination of the rapid motion situation and gradual transition based on another set of features. Apart from the new classification strategy proposed in this paper, imbalanced data classification issue is also taken into consideration, which has not received sufficient attention in previous work. Experimental results show that the new method is effective.