Method and equipment for describing and capturing video object

Abstract

The invention discloses a method and a device for describing and capturing a video object, a video sequence is generated by capturing a video image, a video object tracking sequence is generated according to the video sequence; and a video object descriptor is generated according to the video object tracking sequence and the video sequence. The video object is replayed, and the capture and the tracking are carried out on the video object according to information contained in the obtained video object tracking sequence. The application of the invention does not need the frame-by-frame establishment of the video object descriptor for each video object in that the video object is captured and tracked in the constituted video object tracking sequence according to a tracking serial number of a video object region, thereby reducing the number of the video object descriptors, being well adapted to the application needs of intelligent video interaction and also accelerating the retrieval speed of video information.

Description

technical field [0001] The invention relates to video technology, in particular to a method and device for describing and capturing video objects. Background technique [0002] Video Surveillance (Video Surveillance) technology has gradually formed a network after years of digital development. It is used for production monitoring, such as production line process monitoring, safety monitoring, such as railway station, subway station, airport safety monitoring, and other monitoring such as patient care. However, with the increase of monitoring objects and the update and increase of the attribute information of each object, how to conveniently manage, retrieve, video human-computer interaction, and intelligent reprocessing of a large amount of video surveillance data, and give full play to The advantages of the networked video surveillance system are the problems that the current interactive video system (Interactive Video System) needs to solve. [0003] Moving Picture Expert...

AI Technical Summary

Problems solved by technology

[0009] from figure 1 As can be seen in , since a video object descriptor is established for each object that appears in each frame, each video object descriptor needs to describe the characteristics of the object. Therefore, when the video sequence is long and / or there are many video objects, The number and size of video object descriptors used to describe video objects are very large, which makes it difficult to quickly retrieve video materials

[0010] because figure 1 In the video object descriptor, using contour coordinates to locate the position of the video object requires more complex calculations and takes up more storage resources. Therefore, an improved method is to use the "polygon vertex coordinates" of the video object "or "video object window coordinates" instead of "contour coordinates" to compress the size of the video object descriptor, but use video object window coordinates or polygon vertex coordinates to approximately describe the object contour coordinates. When selecting a video object, if the selected position is inaccurate , it is easy to cause operational errors, and it is rough to track and label video objects using window coordinates or polygon vertex coordinates. Further, this improved method can only reduce the size of video object descriptors, but cannot reduce The number of video object descriptors is still not good for fast retrieval of video materials

Images