实时视频摘要及其在舰船导航中的应用

Real-time Video Abstraction and Its Application in Ship Navigation

针对视频信号在低带宽导航通信条件下进行实时传输比较吃力的问题,本文提出一种新的实时视频摘要算法。该算法是通过修改视觉重要特征(包括亮度、色彩对比等)的对比度来实现。具体地说,在图像的低对比度区域用双边滤波降低对比度,而在高对比度区域人为地加上用高斯差分滤波器得到的边缘以增加对比度。实验结果表明:该算法不仅能够很好地去除噪声,而且具有很强的压缩能力。这在现代导航特别是在低带宽情况下的导航数据传输方面有着巨大的应用前景。

This paper presents a new real-time video and image abstraction method by modifying the contrast of visually important salience including luminance and color opponency. In detail, the paper reduces contrast in low-contrast regions by applying bilateral filter and increase contrast in high-contrast regions with difference-of-Gaussian edges. Experiment results demonstrate that this method not only perfectly removes noises, but also well compresses the images. There is a large application market in modem ship navigation, especially in low-bandwidth data communication.