中文摘要：

对于数字视频镜头突变切换的检测,一般有模板匹配法、直方图法等基本算法,但这些算法都需要确定阈值,并在实际检测中通常达不到较高的检测精度。提出了一种新的基于BP神经网络的视频镜头突变检测算法,该算法选取模板匹配二次差分和直方图二次差分作为特征,利用神经网络的自组织、自学习能力实现镜头突变检测,然后以闪光检测来提高检测的可靠性。实验结果表明,该算法能够有效地检测视频的镜头突变,无须设定阈值,具有计算简单、易于实现的优点。

英文摘要：

There are some basic algorithms, including the template-matching algorithm, the histogram algorithm, etc, used to detect abrupt shot change in digital video, but they all need to fix on thieshold and can＇ t always attain high precision in practical detecting. This paper proposed a new BP neural network based video abrupt shot change detection algorithm. It selected template-matching twice-difference and histogram twice-difference as characters, utilized the neural network＇ s ability of self- organization and self-learning to actualize abrupt shot change detection, then performed the flashlight detection to improve the detecting reliability. Experimental results show that this method achieves satisfying-precision and recall of detecting shot boundaries compared with the conventional schemes, at the same time, it don＇ t need threshold and easy to implement.