Online video concentration device, system and method

The invention discloses an online video concentration device, system and method. The method can sequentially perform real-time execution on an image which is currently acquired by each frame, and the method comprises the following steps of: the segmentation step: segmenting a background image and a foreground image of each image; the extraction step: extracting a motion object from each foreground image; the step of extracting the sequence of the motion objects: accumulating the motion objects which are respectively extracted from the foreground images of the multiple frames and forming the sequence of the motion objects; the step of extracting the sequence of main backgrounds: extracting specific (n) frames of the background images from the background images of the images of the multiple frames as the sequence of the main backgrounds; and the splicing step: splicing the sequence of the main backgrounds with the sequence of the motion objects. By utilizing the online video concentration way, the length of a concentrated video is shortened and information of the motion objects in the video can be retained as far as possible. Fast and convenient video browsing and previewing can be realized, and the visual effect is better. The hardware requirements and the complexity in an algorithm can be reduced.