基于弹药库视频监控的背景提取及阈值获取

武器弹药库安全的重要性不言而喻,历来是部队保卫工作的重点。在军队弹药库视频监控系统中,基于实时场景图像的分割在监控系统中是一个重要环节,因为图像分割使得目标分离、特征提取和参数测量以及对原始图像进行更高层次的分析和语意理解成为可能。作者根据实际情况提出了一个新的获取背景图像的方法和图像分割中阈值的获取算法,对一般迭代法提出改进,并引入一个平滑系数,使分割阈值在原来的水平上得到提高,有效地避免当前帧分割阈值的突变。该算法的鲁棒性好,易于操作,效果明显。

Extracting Background-images and Gaining Valve-value in Video Frequency Monitor System of Ammunition Depot

The importance of ammunition depot security goes without saying, and is always the key point of troop safety security work. In the ammunition depot video frequency monitor system, scene division based on the real-time is an important link, because the scene division makes it possible that goal separation, extracting character, parameter measurement as well as carrying on a higher level analysis to the primitive image and the meaning understands. This system according to the actual situation uses variance and expected value to extract the background image, further more, offers an new algorithm of gaining the valve-value, improves the general iteration process; at the same time introduces a smoothing factor, optimizes the division valve-value, and effectively avoids valve-value mutation. This algorithm is robustness good, and easy to operate, and the effect is obvious.