Retrieval-oriented monitoring video semantic description and inspection modeling method

Abstract

The invention discloses a retrieval-oriented monitoring video semantic description and inspection modeling method. The method comprises the steps as follows: first analyzing the advantages and disadvantages of a present MPEG-7 based video description scheme in a monitoring field, and subdividing video objects into moving objects and stationary regions for description, aiming at the problem that user interested stationary regions are not defined. The method simultaneously defines a second-stage spatial relationship on a time dimension to describe a time variable spatial relationship, in order to solve the problem of description information redundancy in the relationship between objects in a description flow. At last, the method proves that the improved monitoring video semantic description scheme attains a description flow of monitoring video through inspection modeling of events, and can well support the understanding of high-level semantic information.

Description

technical field [0001] The invention relates to the field of description and retrieval of surveillance video, in particular to an improved semantic description and retrieval method for surveillance video. Background technique [0002] The 21st century is an information age. With the rapid development of the information technology industry represented by multimedia, people's requirements for the safety of living and working environments are also increasing, and monitoring facilities are increasingly appearing in various public places. At present, video surveillance systems have been involved in all walks of life in society, such as traffic intersections, supermarkets, banks, stations, and residential quarters. Many monitoring equipment, such as Olympic venues, People's Square, World Expo site, etc. [0003] 1. Problems in traditional video surveillance systems [0004] Video surveillance system is the product of comprehensive application of multimedia technology, network co...

AI Technical Summary

Problems solved by technology

[0005] (1) With the increase in the amount of surveillance probes and video data, its efficiency will be greatly reduced, and it is very difficult to quickly find the required evidence from the massive surveillance video data;

[0006] (2) The transmission and storage of massive video data will cause resource waste and information pollution

[0021] 5. Insufficiency of existing video description schemes

[0025] According to the above analysis, in the video description in the surveillance field, if the static area of ​​interest to the user is not clearly defined and distinguished in the object layer information, it will make it difficult to judge the high-level semantic information later.

[0026] (2) Redundant information describing the relationship between video objects

[0027] Since the spatial relationship between moving objects and between moving objects and static areas changes with time, if the spatial relationship of each frame is described, it will cause redundancy of relationship description information, which is not conducive to mass surveillance video Storage, Transmission and Retrieval of Information

[0028] In addition, some objects do not have a strong correlation, for example, the time or space distance is too large, the description of these relationship information will not help the understanding of high-level semantic information, which also leads to the redundancy of relationship description information

Images