Object-based monitoring video concentration method

Abstract

The invention discloses an object-based monitoring video concentration method. The method comprises the following steps: acquiring a foreground image and a background image of each video frame from the monitoring video, saving the background image in a database, extracting recognition features of the foreground image, orderly comparing the foreground image with feature distances of the recognition features of all foreground images in foreground cache by use of the foreground cache so as to number of objects of all foreground images, wherein, the same object is corresponding to the unique same serial number; and computing a movement direction value and a main color of the same object according to position and color information of the same object, and saving a computation result and the foreground image into the database; finally extracting the foreground image and the background image of the corresponding object from the database, thereby synthesizing a concentrated video. By use of the concentration method disclosed by the invention, the object in different time and spaces can be concentrated, a concentration rate of the monitoring video is improved, and the monitoring efficiency is improved.

Description

technical field [0001] The invention relates to an object-based monitoring video concentration method, which belongs to the technical field of video image processing and pattern recognition. Background technique [0002] With the development of video surveillance technology, in order to reduce the monitoring cost and improve the monitoring efficiency, the surveillance video concentration technology came into being. The video surveillance concentration is to compress the original video in time and space, so that the length of the concentrated video is much shorter than the original video. Video summary, the existing surveillance video enrichment methods mainly include the following two types: [0003] One is the adaptive frame extraction method, which first separates the active foreground in the surveillance video through background modeling, discards or retains specific frames with limited enrichment conditions, such as discarding frames with low activity rates and uninteres...

AI Technical Summary

Problems solved by technology

[0003] One is the adaptive frame extraction method, which first separates the active foreground in the surveillance video through background modeling, discards or retains specific frames with limited enrichment conditions, such as discarding frames with low activity rates and uninteresting frames, and retaining activity rate comparisons High and user-interested frames, some video clips are reserved to represent the video content, and the enrichment conditions can be information such as sound, camera movement, object movement, color, etc. The problem with this method is that it is difficult to meet the needs of all users, Poor enrichment when encountering videos with long active foregrounds

The problem with this method is: when the extracted key frames are too few, the condensed video will lose a lot of important dynamic information in the original video, and when the extracted key frames are too many, a good video concentration effect cannot be achieved

Images